DNA encoding CD40

Abstract

 A simple and highly efficient method for cloning cDNAs from mammalian expression libraries based on transient expression in mammalian host cells has been discovered. Novel expression vectors allowing highly efficient construction of mammalian cDNA libraries are disclosed. The cloning method of the invention which has been used to clone genes for cell surface antigens of human lymphocytes, has general application in gene cloning. Cell surface antigens cloned according to the present invention have been purified, and the nucleotide and amino acid sequences determined. These antigens have diagnostic and therapeutic utility in immune-mediated infections in mammals, including humans.

Description

Background

[0001] A basic tool in the field of recombinant genetics is the conversion of poly(A)⁺ mRNA to double-stranded (ds) cDNA, which then can be inserted into a cloning vector and expressed in an appropriate host cell. Molecular cloning methods for ds cDNA have been reviewed, for example by Williams, "The Preparation and Screening of a cDNA Clone Bank," in Williamson, ed., Genetic Engineer­ing, Vol. 1, p. 2, Academic Press, New York (1981); Maniatis, "Recombinant DNA", in Prescott ed., Cell Biology, Academic Press, New York (1980); and Efstratiadis et al., "Cloning of Double-Stranded DNA," in Stelo et al., Genetic Engineering, Vol. 1, p. 15, Plenum Press, New York (1979).

[0002] A substantial number of variables affect the suc­cessful cloning of a particular gene and cDNA cloning strategy thus must be chosen with care. A method common to many cDNA cloning strategies involves the construction of a "cDNA library" which is a collection of cDNA clones derived from the total poly(A)⁺ mRNA derived from a cell of the organism of interest.

[0003] A mammalian cell may contain up to 30,000 different mRNA sequences, and the number of clones required to obtain low-abundance mRNAs, for example, may be much greater. Methods of constructing genomic eukaryotic DNA libraries in different expression vectors, including bacteriophage λ , cosmids, and viral vectors, are known. Some commonly used methods are described, for example, in Maniatis et al., Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, publisher, Cold Spring Harbor, New York (1982).

[0004] Once a genomic cDNA library has been constructed, it is necessary to isolate from the thousands of host cells the cell containing the particular human gene of interest. Many different methods of isolating target genes from cDNA libraries have been utilized, with varying success. These include, for example, the use of nucleic acid probes, which are labeled mRNA fragments having nucleic acid sequences complementary to the DNA sequence of the target gene. When this method is applied to cDNA clones of abundant mRNAs in transformed bacterial hosts, colonies hybridizing strongly to the probe are likely to contain the target DNA sequences. The identity of the clone then may be proven, for example, by in situ hybridization/selection (Goldberg et al., Methods Enzymol., 68:206 (1979)) hybrid-arrested translation (Paterson et al., Proceedings of the National Academy of Sciences, 74:4370 (1977)), or direct DNA sequencing (Maxam and Gilbert, Proceedings of the National Academy of Sciences, 74:560 (1977); Maat and Smith, Nucleic Acids Res., 5:4537 (1978)).

[0005] Such methods, however, have major drawbacks when the object is to clone mRNAs of relatively low abundance from cDNA libraries. For example, using direct in situ colony hybridization, it is very difficult to detect clones containing cDNA complementary to mRNA species present in the initial library population at less than one part in 200. As a result, various methods for enriching mRNA in the total population (e.g. size fractionation, use of synthetic oligodeoxynucleotides, differential hybridization, or immunopurification) have been developed and are often used when low abundance mRNAs are cloned. Such methods are described, for example, in Maniatis et al., Molecular Cloning: A Laboratory Manual, supra.

[0006] Many functional eukaryotic proteins initially exist in the form of precursor molecules which contain leader or signal sequences at their N-terminal ends. These leader sequences bind to the cell membrane and draw the remainder of the protein through the lipid bilayer, after which the signal sequence is cleaved from the protein by a signal peptidase enzyme. The protein thus functions only after secretion from the cells (for example, insulin, serum albumin, antibodies, and digestive tract enzymes), or after the proteins have been anchored to the outer surface of a cell membrane (for example, histocom­patibility antigens).

[0007] The cell surface antigens characteristic of mammalian T lymphocytes are additional examples of proteins that anchor to the cell surface. In mammals, certain cells derived from bone marrow mature into lymphocytes, which are present in the lymphoid organs, including the thymus, spleen, lymph nodes, and lymphoid aggregates, and also circulate actively through the blood and lymph systems. Mature lymphocyte cells may be divided into two populations: thymus-dependent (T) lymphocytes and thymus-independent (B) lymphocytes. T lymphocytes migrate to the interior of the thymus, where they undergo differentiative proliferation. During their differentiation process, they express characteristic cell surface membrane alloantigens, including Thy-1, TLA, gv-1, Ly-1, Ly-2, Ly-3, and Ly-5. As they mature, T lymphocytes lose the TLA antigens and some of the Thy-1 antigens, and gain histocompatibility antigens, acquiring the membrane conformation typical of the recirculating T lymphocytes. This is described, for example, by Mota, "Activity of Immune Cells," in Bier et al., eds., Fundamentals of Immunology, 2d Ed., Springer-Verlag, Berlin, pp. 35-62 (1986).

[0008] T lymphocytes are involved indirectly in the formation of antibodies and their activities thus have required complex analysis of cell function, rather than simple antibody titer measurement. Partly due to this, their importance in development of immunologic competence was not recognized until relatively recently. Mature T lymphocytes synthesize and express an unique pattern of surface glycoprotein antigens which serve as markers for identification of different T lymphocyte subpopulations, including T helper cells, T suppressor cells, and T cytotoxic cells. Each of these subpopulations plays a very important role in regulating the immune system. (Mota, supra).

[0009] In humans, the functional and phenotypic heterogene­ity of T lymphocytes is well accepted. Two major subpop­ulations are known: effector T cells mediating cellular immunity; and regulator T cells containing helper and suppressor T lymphocytes. These two subpopulations have been defined with heteroantisera, autoantibodies, and monoclonal antibodies directed at cell surface antigens. For example, earlier in their development, human lymphoid cells in the thymus express an antigen designated T11 which reacts strongly to a monoclonal antibody designated Cluster of Differentiation 2 (CD2), and react slightly with monoclonal antibody CD5 to cell surface antigen T1. During maturation, these cells lose T11 (CD2) and acquire three new antigens defined by monoclonal antibodies CD4, CD8, and CD1. With further maturation, the thymocytes cease to express cell surface antigens reactive with monoclonal antibody CD1, express the T3 antigen reactive with monoclonal antibody CD3, and then segregate into two subpopulations which express either T4 (CD4) or T8 (CD8) antigen. Immunologic competence is acquired at this stage, but is not completely developed until thymic lymphocytes migrate outside the thymus. (Mota, supra.) In contrast with the majority of thymocytes, circulating T lymphocytes express the T1 (CD5) and T3 (CD3) antigens. The T4 (CD4) antigen is present on approximately 55-65% of peripheral T lymphocytes, whereas the T8 (CD8) antigen is expressed on 20-30%. These two subpopulations corre­spond to helper and to suppressor and cytotoxic T cells, respectively.

[0010] In addition to providing a convenient means of distinguishing T lymphocyte subpopulations, these cell surface antigens are important for mature T cell ac­tivation and effector function. T cell activation involves a complex series of cell surface interactions between the T cell and the target cell or stimulator cell in addition to binding of the T cell receptor to its specific antigen.

[0011] For example, CD2, the human T cell erythrocyte receptor, allows thymocytes and T-lymphocytes to adhere to target cells (e.g., erythrocytes) and to thymic epithelium. This occurs via a specific molecular ligand for CD2, designated LFA-3, in humans, which is a widely distributed surface antigen. This phenomenon has long been employed to detect, assay and purify human cells producing antibodies to sheep erythrocytes and serves as the basis for the E-rosette test, first described by Zaalberg, Nature, 202:1231 (1964). CD2/LFA-3 interactions also have been shown to mediate cytolytic target conjugation (Shaw et al., Nature, 323:262-264 (1986), and the mixed lymphocyte reaction (Martin et al., J. Immunol., 131:180-185 (1983). Anti-CD2 monoclonal antibodies can directly activate peripheral T-lymphocytes via an antigen-independent pathway (Meuer et al., Cell, 36:897-906 (1984), indicating an even wider immunoregulatory role for CD2.

[0012] Recognition that T lymphocytes are the main effectors of cell-mediated immunity and also are involved as helper or suppressor cells in modulating the immune response has resulted in a significant contribution to the increasing practical application of clinical immunology to medicine. The scope of this application includes defense against infections, prevention of diseases by immunization, organ transplantation, blood banking, deficiencies of the immune system, and a variety of disorders that are mediated by immunologic mechanisms. Moreover, immunologic techniques frequently are used in the clinical laboratory, as in the measurement of hor­mones and drugs. Clinical immunology is described, for example, in Weir, ed., Handbook of Experimental Immunology in Four Volumes: Volume 4: Applications of Immunological Methods in Biomedical Sciences, 4th Ed., Blackwell Scientific Publications, Oxford (1986); Boguslaski et al., eds., Clinical Immunochemistry: Principles of Methods and Applications, Little, Brown & Co., Boston (1984); Holborow et al., eds., Immunology in Medicine: A Comprehensive Guide to Clinical Immunology, 2d Ed., Grune & Stratton, London (1983); and Petersdorf et al., eds., Harrison's Principles of Internal Medicine, 10th ed., McGraw-Hill, New York, publisher, pp. 344-391 (1983). Clearly, a more thorough understanding of the proteins which mediate the immune system would be of significant value in clinical immunology.

[0013] Use of mammalian expression libraries to isolate cDNAs encoding mammalian proteins such as those described above would offer several advantages. For example, the protein expressed in a mammalian host cell should be functional and should undergo any normal posttranslational modification. A protein ordinarily transported through the intracellular membrane system to the cell surface should undergo the complete transport process. A mammalian expression system also would allow the study of intracellular transport mechanisms and of the mechanism that insert and anchor cell surface proteins to membranes.

[0014] One common mammalian host cell, called a "COS" cell, is formed by infecting monkey kidney cells with a mutant viral vector, designated simian virus strain 40 (SV40), which has functional early and late genes, but lacks a functional origin of replication. In COS cells, any foreign DNA cloned on a vector containing the SV40 origin of replication will replicate because SV40 T antigen is present in COS cells. The foreign DNA will replicate transiently, independently of the cellular DNA.

[0015] With the exception of some recent lymphokine cDNAs isolated by expression in COS cells (Wong, G.G., et al., Science, 228:810-815 (1985); Lee, F. et al., Proceedings of the National Academy of Sciences, USA, 83:2061-2065 (1986); Yokota, T., et al., Proceedings of the National Academy of Sciences, USA, 83:5894-5898 (1986); Yang, Y., et al., Cell, 47:3-10 (1986)), however, few cDNAs in general are isolated from mammalian expression libraries. There appear to be two principal reasons for this: First, the existing technology (Okayama, H. et al., Mol. Cell. Biol., 2:161-170 (1982)) for construction of large plasmid libraries is difficult to master, and library size rarely approaches that accessible by phage cloning techniques. (Huynh, T. et al., In: DNA Cloning Vol. I, A Practical Approach, Glover, D.M. (ed.), IRL Press, Oxford (1985), pp. 49-78). Second, the existing vectors are, with one exception (Wong G.G., et al., Science, 228:810-815 (1985)), poorly adapted for high level expression, particularly in COS cells. The reported successes with lymphokine cDNAs do not imply a general fitness of the methods used, since these cDNAs are particularly easy to isolate from expression libraries. Lymphokine bioassays are very sensitive ((Wong, G.G., et al., Science, 228:810-815 (1985); Lee, F. et al., Proceedings of the National Academy of Sciences, USA, 83:2061-2065 (1986); Yokota T. et al., Proceedings of the National Academy of Sciences, USA, 83:5894-5898 (1986); Yang, Y. et al., Cell, 47:3-10 (1986)) and the mRNAs are typically both abundant and short (Wong, G.G. et al., Science, 228:810-815 (1985); Lee, F., et al., Proceedings of the National Academy of Sciences, USA, 83:2061-2065 (1986); Yokota, T., et al., Proceedings of the National Academy of Sciences, USA, 83:5894-5898 (1986); Yang, Y., et al., Cell, 47:3-10 (1986)).

[0016] Thus, expression in mammalian hosts previously has been most frequently employed solely as a means of verifying the identity of the protein encoded by a gene isolated by more traditional cloning methods. For example, Stuve et al., J. Virol., 61(2):327-335 (1987), cloned the gene for glycoprotein gB2 of herpes simplex type II strain 333 by plaque hybridization of M13-based recombinant phage vectors used to transform competent E. coli JM101. The identity of the protein encoded by the clone thus isolated was verified by transfection of mammalian COS and Chinese hamster ovary (CHO) cells. Expression was demonstrated by immunofluorescence and radioimmunoprecipitation.

[0017] Oshima et al. used plaque hybridization to screen a phase lambdagt11 cDNA library for the gene encoding human placental beta-glucuronidase. Oshima et al., Proceedings of the National Academy of Sciences, U.S.A., 84:685-689 (1987). The identity of isolated cDNA clones was verified by immunoprecipitation of the protein expressed by COS-7 cells transfected with cloned inserts using the SV40 late promoter.

[0018] Transient expression in mammalian cells has been employed as a means of confirming the identity of genes previously isolated by other screening methods. Gerald et al., Journal of General Virology, 67:2695-2703(1986). Mackenzie, Journal of Biological Chemistry, 261:14112-­14117 (1986); Seif et al., Gene, 43:1111-1121 (1986); Orkin et al., Molecular and Cellular Biology, 5(4):762-­767 (1985). These methods often are inefficient and tedious and require multiple rounds of screening to identify full-length or overlapping clones. Prior screening methods based upon expression of fusion proteins are inefficient and require large quantities of monoclonal antibodies. Such drawbacks are compounded by use of inefficient expression vectors, which result in protein expression levels that are inadequate to enable efficient selection.

Summary of the Invention

[0019] The present invention relates to a powerful new method for cloning cDNA encoding cell surface antigens, to a method of construction cDNA libraries to high efficiency expression vectors particularly suited for high level expression in eukaryotic host cells, and to the isolated nucleotide sequences and their encoded products.

[0020] The highly efficient cloning technique of the present invention is based upon transient expression of antigen in eukaryotic cells and physical selection of cells expressing the antigen by adhesion to an antibody-coated substrate, such as a culture dish. The methods of the present invention are useful for the isolation and molecular cloning of any protein which can be expressed and transported to the cell surface membrane of a eukaryotic cell.

[0021] The method for cloning cDNA encoding a cell surface antigen of the present invention comprises preparing a cDNA library; introducing this cDNA library into eukaryotic mammalian preferably tissue culture cells; culturing these cells under conditions allowing expression of the cell surface antigen; exposing the cells to a first antibody or antibodies directed against the cell surface antigen, thereby allowing the formation of a cell surface antigen-first antibody complex; subsequently exposing the cells to a substrate coated with a second antibody directed against the first antibody, thereby causing cells expressing the cell surface antigen to adhere to the substrate via the formation of a cell surface antigen-first antibody-second antibody complex; and separating adherent from non-adherent cells.

[0022] By means of the cloning method of the present invention, isolation and molecular cloning of genes encoding such cell surface antigens as the following have been accomplished: the CD1a, CD1b, CD1c, CD2, CD6, CD7, CD13, CD14, CD16, CD19, CD20, CD22, CD26, CD27, CD28, CD31, CDw32a, CDw32b, CD33, CD34, CD36, CD37, CD38, CD39, CD40, CD43, ICAM, LFA-3, FcRIa, FcRIb, TLiSa, and Lev-8 antigens. The nucleotide sequences of genes cloned by the method of the present invention have been determined and the amino acid sequences of the encoded proteins have been identified. A cloned gene, such as that encoding CD1a, CD1b, CD1c, CD2, CD6, CD7, CD13, CD14, CD16, CD19, CD20, CD22, CD26, CD27, CD28, CD31, CDw32a, CDw32b, CD33, CD34, CD36, CD37, CD38, CD39, CD40, CD43, ICAM, LFA-3, FcRIa, FcRIb, TLiSa, and Lev-8, is also the subject of the present invention.

[0023] Once the gene encoding an antigen has been cloned according to the method of the present invention, that gene can be expressed in a prokaryotic or a eukaryotic host cell to produce the encoded protein or portion thereof in substantially pure form such as it does not exist in nature. Another aspect of the present invention relates to substantially pure cell surface antigens, particularly: CD1a, CD1b, CD2, CD6, CD7, CD13, CD14, CD16, CD19, CD20, CD22, CD26, CD27, CD28, CD31, CDw32a, CDw32b, CD33, CD34, CD36, CD37, CD38, CD39, CD40, CD43, ICAM, LFA-3, FcRIa, FcRIb, TLiSa, and Lev-8 antigens and their functional analogues and equivalents. The primary amino acid sequences of the CD1a, CD1b, CD2, CD7, CD14, CD16, CD19, CD20, CD22, CD28, CDw32a, CDw32b, CD33, CD34, CD40, ICAM, LFA-3, FcRIa, FcRIb and TLiSa antigens have been determined. The invention thus also relates to the amino acid sequences of those antigens and their function equivalents and to the nucleotide sequences encoding those antigens.

[0024] This invention also relates to high efficiency cDNA expression vectors which allow the generation of very large mammalian expression libraries and yield large amounts of protein in mammalian host cells, resulting in efficient selection. In a particular embodiment of this invention, a cDNA expression vector comprises a sup­pressor tRNA gene; an SV40 origin; a synthetic transcrip­tion unit, comprising a chimeric promotor composed of human cytomegalovirus AD169 immediate early enhancer sequencer fused to the HIV LTR -60 to +80 sequences, inserted between the suppressor tRNA gene and the SV40 origin; a polylinker comprising two BstXI sites separated by a replaceable DNA sequence and flanked by XbaI sites; and an SV40 small t antigen splice and early region polyadenylation signals.

[0025] A further aspect of the present invention comprises a synthetic transcription unit for use in a cDNA expres­sion vector, comprising a chimeric promoter composed of human cytomegalovirus AD169 immediate early enhancer sequences fused to HIV LTR -60 to +80 sequences. The small size and particular arrangement of the sequences of the cDNA expression vector of the present invention allow highly efficient replication in host mammalian tissue culture cells, such as COS cells. Moreover, this vector employs a polylinker containing two inverted BstXI sites separated by a short replaceable DNA segment, which allows the use of a very efficient oligonucleotide-based cDNA insertion strategy.

[0026] In another aspect, the present invention comprises a vector comprising two identical BstXI sites in inverted orientation each with respect to the other, which BstXI sites are separated by a short replaceable DNA fragment. Another aspect of the invention is a polylinker as described above.

[0027] A further aspect of the invention relates to an oligonucleotide-based cDNA insertion method, comprising ligating synthetic DNA oligonucleotides to the cDNA segment desired to be inserted into a vector, the syn­thetic DNA oligonucleotides giving the same terminal sequences as those of the short replaceable DNA fragment of the polylinker of the invention, and inserting the resulting cDNA segment plus synthetic DNA oligonucleotide terminal sequences into the polylinker of the vector, from which the short replaceable DNA fragment previously has been removed.

[0028] In preparing cDNA libraries according to the present invention, it has been discovered that many tumors are heavily infiltrated by macrophages and lymphocytes, and thus may be employed as a source of macrophage or lympho­cyte transcripts to good effect, instead of tumor cell lines commonly used. In another aspect, then, the present invention relates to the use of tumor cells, particularly human tumor cells, to prepare cDNA libraries for use according to the methods of the present inven­tion.

[0029] Another advantage of the powerful selection system of the present invention is that directional insertion of the cDNA is not necessary. The method of the present invention results in library construction efficiencies which are on a par with those described for phage vectors such as λ gt10 and λ gt11, with the additional advantage that clones generated according to the methods of the present invention are easier to manipulate.

[0030] The immunoselection technique of the present inven­tion allows efficient use of antibodies, which may be monoclonal or polyclonal, in relatively small absolute amounts. The method of the present invention also is quite rapid. Generally, three or fewer cycles of immuno­selection and rescue are required to isolate a target cDNA clone. Thus, the method of the present invention also results in the efficient use of labor and materials when cloning genes encoding cell surface antigens. As described above, this method has been employed to successfully clone genes encoding cell surface antigens associated with mammalian T lymphocytes (e.g. antigens CD1a, CD1b, CD1c, CD2, CD6, CD7, CD13, CD14, CD16, CD19, CD20, CD22, CD26, CD27, CD28, CD31, CDw32a, CDw32b, CD33, CD34, CD36, CD37, CD38, CD39, CD40, CD43, ICAM, LFA-3, FcRIa, FcRIb, TLiSa, and Lev-8).

[0031] The purified genes and proteins of the present invention are useful for immunodiagnostic and immuno­therapeutic applications, including the diagnosis and treatment of immune-mediated infections, diseases, and disorders in animals, including humans. They can also be used to identify, isolate and purify other antibodies and antigens. Such diagnostic and therapeutic uses comprise yet another aspect of the present invention. Moreover, the substantially pure proteins of the present invention may be prepared as medicaments or pharmaceutical compositions for therapeutic administration. The present invention further relates to such medicaments and compositions.

Brief Description of the Drawings

Figure 1. Nucleotide sequence of expression vector piH3

[0032] Nucleotides 1-589 are derived from pMB1 origin (pBR322 ori); nucleotides 590-597 are derived from the SacII linker (ACCGCGT); nucleotides 598-799 are derived from the synthetic tyrosine suppressor tRNA gene (supF gene); nucleotides 800-947 are derived from a remnant of the ASV LTR fragment (PvuII to MIu1); nucleotides 948-1500 are derived from the human cytomegalovirus AD169 enhancer, nucleotides 1501-1650 are derived from HIV TATA and tat-responsive elements; nucleotides 1651-1716 are derived from the piLNXAN polylinker (HindIII to Xba); nucleotides 1717-2569 are derived from pSV to splice and poly-Addition signals; nucleotides 2570-2917 are derived from the SV40 origin of replication (pvuII to (HindIII); and nucleotides 2918-2922 are derived from piVX, remnant of R1 site from polylinker.

Figure 2. Nucleotide sequence of the CD2cDNA insert

[0033] Nucleotide numbering is given in parentheses at right, amino acid numbering, center and left. Locations of the potential sites for addition of asparagine-linked carbohydrate (CHO) are shown, as well as the predicted transmembrane (TM) sequence. The amino acid sequence is numbered from the projected cleavage site of the secretory signal sequence. Proline residues are underscored with asterisks. The presumed polyadenylation signal for the 1.65kb transcript, and a possible polyadenylation signal for the 1.3kb transcript, are shown boxed.

Figure 3. Restriction map of the CDM8 expression vector

[0034] The CDM8 vector includes a deleted version of a mutant polyoma virus early region selected for high efficiency expression in both murine and monkey cells. Substantially all of the human immunodeficiency promoter region has been replaced with the cognate sequences of the human cytomegalovirus immediate early promoter, and by inclusion of a bacteriophage T7 promoter between the eukaryotic promoter and the site of cDNA insertion. Arrows indicate the direction of transcription.

Figure 4. Nucleotide sequence and corresponding amino acid sequence of the LFA-3 antigen

[0035] WOP cells transfected with a clone encoding the LFA-3 antigen were detected by indirect immuno­fluorescence, amplified and sequenced. (A) shows the 874 base pair insert containing an open reading frame of 237 residues originating at a methionine codon, and terminating in a series of hydrophobic residues. Hydrophobic and hydrophilic regions within this open reading frame are shown in (B).

Figure 5. Restriction Map of the piH3M vector

[0036] The direction of transcription is indicated by an arrow. Restriction endonuclease sites flanking the BstXI cloning sites are shown.

Figure 6. Nucleotide sequence of the piH3M vector

[0037] There are 7 segments. Residues 1-587 are from the pBR322 origin of replication, 588-1182 from the M13 origin, 1183-1384 from the supF gene, 1385-2238 are from the chimeric cytomegalovirus/human immunodeficiency virus promoter, 2239-2647 are from the replaceable fragment, 2648-3547 from plasmid pSV2 (splice and polyadenylation signals), and 3548-3900 from the SV40 virus origin.

Figure 7. Nucleotide sequence of the CD28 cDNA

[0038] Nucleotide numbering is given in parentheses at right, amino acid numbering, center and left. Location of the potential sites for addition of asparagine-linked carbohydrate (CHO) are shown, as well as the predicted transmembrane (TM) sequence. The amino acid sequence is numbered from the projected cleavage site of the secretory signal sequence. The presumed polyadenylation signal is shown boxed.

Figure 8. Nucleotide sequence of the CD7 cDNA insert

[0039] Nucleotide numbering is given in parentheses at right. The amino acid sequence is numbered from the projected cleavage site of the secretory signal sequence. Splice donor and acceptor sites indicated by (/). The location of the potential sites for addition of asparagine-linked carbohydrate (CHO) are shown, the potential fatty acid esterification site is denoted (*), and the predicted transmembrane domain (TM) is underlined. Nucleotide sequences potentially involved in hairpin formation are denoted by (.). The presumed polyadenylation signal is underlined.

Figure 9. Nucleotide sequence of the CDw32 cDNA

[0040] Nucleotide number is given in the parenthesis at right, amino acid numbering, center and left. Locations of the potential sites for addition of asparagine-linked carbohydrate (CHO) are shown, as well as the predicted transmembrane (TM) sequence. The amino acid sequence is numbered from the projected cleavage site of the secretory signal sequence. Cysteine residues are underscored with asterisks.

Figure 10. Sequence of the CD20.4 cDNA

[0041] 

A. The sites of potentional N-linked glycosylation are denoted by the symbol -CHO-: the hydrophobic regions are underscored. The site of the poly(A)⁺ tail in clone CD20.6 is denoted by an asterisk.

B. Hydrophobicity profile of the amino acid sequence in A.

Figure 11. Sequence of ICAM-1

[0042] Complete nucleotide sequence of ICAM-1 cDNA insert and predicted protein sequence. Nucleotide numbering is at right, amino acid numbering in parentheses at left. The RGE motif at position 152 is underlined, the potentional N-linked glycosylation sites are indicated by -CHO- and the transmembrane domain by -TM-. The amino acid sequence is numbered from the projected cleavage site of the signal peptide. Sequencing was by dideoxy-chain termination (Sanger, F., et al., Proc. Natl. Acad. Sci. USA, 74:5463-5467 (1977)), using a combination of subclones, and specific oligonucleotides.

Figure 12. Nucleotide sequence of CD19

[0043] 

Figure 13. Nucleotide sequence of CD20

Figure 14. Nucleotide sequence of CD22

Figure 15. Nucleotide sequence of CDw32a

Figure 16. Nucleotide sequence of CDw32b

Figure 17. Nucleotide sequence of CD40

Detailed Description of the Invention

[0044] This invention relates to a novel method for cloning cDNA encoding a cell surface antigen and to a method of constructing cDNA libraries. It also relates to particular cDNA expression vectors and components thereof, nucleotide sequences or genes isolated by the method, substantially pure cell surface antigens encoded by the cDNA segments, and methods of using the isolated nucleotide sequences and encoded products.

[0045] In the following description, reference will be made to various methodologies known to those of skill in the art of recombinant genetics. Publications and other materials setting forth such known methodologies to which reference is made are incorporated herein by reference in their entireties.

[0046] Standard reference works setting forth the general principles of recombinant DNA technology include Darnell, J.E. et al., Molecular Cell Biology, Scientific American Books, Inc., publisher, New York, N.Y. (1986); Lewin, B.M., Genes II, John Wiley & Sons, publisher, New York, N.Y. (1985); Old, R.W. et al., Principles of Gene Manipulation: An Introduction to Genetic Engineering, 2d edition, University of California Press, Berkeley, CA (1981); and Maniatis, T. et al., Molecular Cloning: A Laboratory Manual, Cold Spring Harbor, NY (1982).

[0047] By "cloning" is meant the use of in vitro recom­bination techniques to insert a particular gene or other DNA sequence into a vector molecule. In order to suc­cessfully clone a desired gene, it is necessary to employ methods for generating DNA fragments, for joining the fragments to vector molecules, for introducing the composite DNA molecule into a host cell in which it can replicate, and for selecting the clone having the target gene from amongst the recipient host cells.

[0048] By "cDNA" is meant complementary or copy DNA produced from an RNA template by the action of RNA-dependent DNA polymerase (reverse transcriptase). Thus a "cDNA clone" means a duplex DNA sequence comple­mentary to an RNA molecule of interest, carried in a cloning vector.

[0049] By "cDNA library" is meant a collection of recombinant DNA molecules containing cDNA inserts which together comprise the entire genome of an organism. Such a cDNA library may be prepared by art-recognized methods described, for example, in Maniatis et al., Molecular Cloning: A Laboratory Manual, supra. Generally, RNA is first isolated from the cells of an organism from whose genome it is desired to clone a particular gene. Preferred for the purposes of the present invention are mammalian, and particularly human, cell lines. More preferred are the human tumor cell line HPB-ALL and the human lymphoblastoid cell line JY. Alternatively, RNA can be isolated from a tumor cell, derived from an animal tumor, and preferably from a human tumor. Thus, a library may be prepared from, for example, a human adrenal tumor, but any tumor may be used.

[0050] The immunoselection cloning method of the present invention comprises the preparation of a cDNA library by extracting total RNA including a particular gene from a cell, synthesizing a series of complementary double-stranded cDNA fragments from the RNA and introducing these cDNA fragments into mammalian cells in tissue culture. The mammalian cells are maintained under conditions which allow them to express the protein (i.e. the cell surface antigen). The resulting cells are exposed to a first antibody or pool (group) of antibodies directed against the cell surface antigen. This results in formation of a cell surface antigen-first antibody complex. The complexes are exposed to a substrate to which is coated or bound a second antibody directed against the first antibody. Cells expressing the cell surface antigen adhere to the substrate (because of formation of a cell surface antigen-first antibody-second antibody complex). Adherent cells are separated from non-adherent cells.

Isolation of total RNA

[0051] The guanidium thiocyanate/CsCl method of isolating total RNA is preferred. More preferred is a guanidium thiocyanate/LiCl variant of the GuSCN/CsCl method, which has added capacity and speed. Briefly, for each ml of mix desired, 0.5g GuSCN are dissolved in 0.58ml of 25% LiCl (stock filtered through 0.45 micron filter) and 20ul of mercaptoethanol is added. Cells are spun out and the pellet is dispersed on walls by flicking, add 1ml of solution to up to 5 x 10⁷ cells. The resulting combination is sheared by polytron until nonviscous. For small scale preps (less than 10⁸ cells) layer 2ml of sheared mix on 1.5ml of 5.7M CsCl (RNase free; 1.26 g CsCl added to every ml of 10mM EDTA pH 8), overlay with RNase-free water and spin SW55 50krpm 2h. For large scale preps, layer 25ml of 12ml CsCl in a SW28 tube, overlay, and spin 24k rpm 8h. Aspirate contents carefully with a sterile pasteur pipet connected to a vacuum flask. Once past the CsCl interface, scratch a band around the tube with the pipet tip to prevent the layer on the wall of the tube from creeping down. The remaining CsCl solution is aspirated. The pellets are taken up in water (do not try to redissolve). 1/10 vol. NaOAc and 3 vol. EtOH are added and the resulting combination is spun. If necessary, the pellet is resuspended in water (e.g., at 70°). Adjust concentration to 1mg/ml and freeze. Small RNA (e.g. 5S) does not come down. For small amounts of cells, scale down volumes and overlay GuSCN with RNase-free water on gradient (precipitation is inefficient when RNA is dilute).

Preparation of poly A⁺ RNA

[0052] Next, polyA⁺ RNA may be prepared, preferably by the oligo dT selection method. Briefly, a disposable polypropylene column is prepared by washing with 5M NaOH and then rinsing with RNase-free water. For each milli­gram total RNA about 0.3 ml (final packed bed) oligo dT cellulose is used. Oligo dT cellulose is prepared by resuspending about 0.5 ml of dry powder in 1 ml of 0.1M NaOH and transferring it into the column, or by percolating 0.1 NaOH through a previously used column (columns can be reused many times). This is washed with several column volumes of RNase-free water, until pH is neutral, and rinsed with 2-3 ml of loading buffer. The column bed is then removed into a sterile 15ml tube using 4-6 ml of loading buffer. The total RNA to 70°C for 2-3 min., LiCl from RNase-free stock is added (to 0.5M), and combined with oligo dT cellulose in a 15 ml tube. This is followed by vortexing or agitation for 10 min. The result is poured into a column and washed with 3 ml loading buffer and then 3 ml of middle wash buffer. mRNA is eluted directly into an SW55 tube with 1.5 ml of 2mM EDTA, 0.1% SDS; the first two or three drops are discarded.

[0053] Eluted mRNA is precipitated by adding 1/10 vol. 3M NaOAc and filling the tube with EtOH. This is then mixed, chilled for 30 minutes at -20°C, and spun at 50k rpm at 5°C for 30 min. The EtOH is poured off and the tube is air dried. The mRNA pellet is resuspended in 50-100ul of RNase-free water. Approximately 5 ul is melted at 70° in MOPS/EDTA/formaldehyde and run on an RNase-free 1% agarose gel to check quality.

cDNA Synthesis

[0054] From this, cDNA is synthesized. A preferred method of cDNA synthesis is a variant of that described by Gubler and Hoffman, (Gene 25:263-269 (1982)). This is carried out as follows:

a. First Strand. 4 ug of mRNA and heated to about 100°C in a microfuge tube for 30 seconds and quenched on ice. The volume is adjusted to 70 ul with RNase-free water. The following are added: 20 ul of RT1 buffer, 2 ul of RNAse inhibitor (Boehringer 36 u/ul), 1 ul of 5 ug/ul of oligo dT (Collaborative Research), 2.5 ul of 20 mM dXTP's (ultrapure), 1 ul of 1 M DTT and 4 ul of RT-LX (Life Science, 24 u/ul). The resulting combination is incubated at 42°C for 40 min. It is heated to inactivate (70°C 10 min).

b. Second Strand. 320 ul of RNAse free water, 80 ul of RT2 buffer, 5 ul of DNA Polymerase I (Boehringer, 5 U/ul), 2 ul RNAse H (BRL 2 u/ul). Incu­bate at 15°C for 1 hr and 22°C for 1 hr. Add 20 ul of 0.5M EDTA pH 8.0, phenol extract and EtOH precipitate by adding NaCl to 0.5M, linear polyacrylamide (carrier) to 20ug/ml, and filling tube with EtOH. Spin 2-3 minutes in microfuge, remove, vortex to dislodge precipitate high up on wall of tube, and respin 1 minute.

c. Adaptors. Resuspend precipitated cDNA in 240 ul of TE (10/1). Add 30 ul of 10x low salt buffer, 30ul of 10X low salt buffer, 30ul of 10X ligation additions, 3ul (2.4ug) of kinased 12-mer adaptor, 2ul (1.6ug) of kinased 8-mer adaptor, and 1 ul of T4 DNA ligase (BioLabs, 400 u/ul, or Boehringer, 1 Weiss unit/ml). Incubate at 15°C overnight. Phenol extract and EtOH precipitate as above (no extra carrier now needed), and resuspend in 100 ul of TE.

Use of cDNA fragments in expression vectors

[0055] For use with the BstXI-based cDNA expression vectors of the invention, (see infra) oligonucleotide segments containing terminal sequences corresponding to BstXI sites on the vectors are ligated to the cDNA fragment desired to be inserted. The resulting fragments are pooled by fractionation. A preferred method is as follows:

[0056] Prepare a 20% KOA, 2mM EDTA, 1 ug/ml EthBr solution and a 5% KOAc, 2mM EDTA, 1 ug/mlg EthBr solution. Add 2.6 ml of 20% KOAc solution to back chamber of a small gradient maker. Remove air bubble from tube connecting the two chambers by allowing solution to flow into the front chamber and then tilt back. Close passage between chambers, and add 2.5ml. of the 5% solution to the front chamber. If there is liquid in the tubing from a previous run, allow the 5% solution to run just to the end of the tubing, and then return to chamber. Place the apparatus on a stirplate, set the stir bar moving as fast as possible, open the stopcock connecting the two cham­bers and then open the front stopcock. Fill a polyallomer SW55 tube from the bottom with the KOAc solution. Overlay the gradient with 100 ul of cDNA solution. Prepare a balance tube and spin the gradient for 3 hrs at 50k rpm at 22°C. To collect fractions from the gradient, pierce the SW55 tube with a butterfly infusion set (with the luer hub clipped off) close to the bottom of the tube and collect three 0.5ml fractions and then 6 0.25ml fractions into microfuge tubes (about 22 and 11 drops respectively). EtOH precipitate the fractions by adding linear polyacrylamide to 20 ug/ml and filling the tube to the top with EtOH. After cooling tubes, spin them in a microfuge for 3 min. Vortex and respin 1 min. Rinse pellets with 70% EtOH (respin). Do not dry to completion. Resuspend each 0.25ml fraction in 10 ul of TE. Run 1 ul on a 1% agarose minigel. Pool the first three fractions, and those of the last six which contain no material smaller than 1kb.

[0057] Suppressor tRNA plasmids may be propagated by known methods. In a preferred method according to the present invention, supF plasmids can be selected in nonsuppress­ing hosts containing a second plasmid, p3, which contains amber mutated ampicillin and tetracycline drug resistance elements (Seed, 1983). The p3 plasmid is derived from PR1, is 57kb in length, and is a stably maintained, single copy episome. The ampicillin resistance of this plasmid reverts at a high rate, so that amp^r plasmids usually cannot be used in p3-containing strains. Se­lection for tet resistance alone is almost as good as selection for ammp+tet resistance. However, spontaneous appearance of chromosomal suppressor tRNA mutations presents an unavoidable background (frequency about 10⁻⁹) in this system. Colonies arising from spontaneous suppressor mutations are usually bigger than colonies arising from plasmid transformation. Suppressor plasmids typically are selected for in LB medium containing amp at 12.5 ug/ml and tet at 7.5 ug/ml. For large plasmid preps, M9 casamino acids medium containing glycerol (0.8%) may be used as a carbon source, and the bacteria grown to saturation.

[0058] Vector DNA may be isolated by known methods. The following method is preferred for plasmid from 1 liter of saturated cells:

[0059] Spin down cells in 1 liter J6 bottles, 4.2k rpm, 25 minutes. Resuspend in 40 ml 10mM EDTA pH 8 (Thump on soft surface). Add 80 ml 0.2M NaOH, 1% SDS, swirl until clearish, viscous. Add 40 ml 5M KOAc, pH4.7 (2.5M KOAc, 2.5M HOAc) shake semi-vigorously (until lumps are 2-3 mm in size). Spin (same bottle) 4.2 rpm, 5 min. Pour supernatant through cheesecloth into 250 ml bottle. Fill bottle with isopropyl alcohol. Spin J6, 4.2k rpm, 5 min. Drain bottle, rinse gently with 70% EtOH (avoid fragmenting the pellet). Invert bottle, and remove traces of EtOH with Kimwipe. Resuspend in 3.5 ml Tris base/EDTA 20mM/10mM. Add 3.75 ml of resuspended pellet to 4.5g CsCl. Add 0.75 ml 10/mg/ml ethidium bromide, mix. Fill VTi80 tubes with solution. Run at a speed of 80 rpm for 2.5 hours or longer. Extract bands by visible light with 1 ml syringe and 20 gauge or lower needle. Cut top off tube, insert the needle upwards into the tube at an angle of about 30° with respect to the tube, (i.e., as shallowly as possible) at a position about 3mm beneath the band, with the bevel of the needle up. After the band is removed, pour tube contents into bleach. Deposit extracted bands in 13 ml Sarstedt tube. Fill tube to top with n-butanol saturated with 1M NaCl, extract. If a very large quantity of DNA is obtained, reextract. Aspirate butanol into trap containing 5M NaOH (to destroy ethidium). Add about equal volume 1M ammonium acetate to DNA (squirt bottle). Add about 2 volumes 95% ethanol (squirt bottle). Spin 10K rpm, 5 min. J2-21. Rinse pellet carefully with 70% ethanol. Dry with swab, or lyophilizer.

[0060] The vector may be prepared for cloning by known methods. A preferred method begins with cutting 20 ug of vector in a 200 ul reaction with 100 units of BstXI (New York Biolabs), cutting at 50°C overnight in a well-thermostatted water bath (i.e., circulating water bath). Prepare 2 KOAc 5-20% gradients in SW55 tubes as described above. Add 100 ul of the digested vector to each tube and run for 3 hrs, 50K rpm at 22°C. Examine the tube under 300nm UV light. The desired band will have migrated 2/3 of the length of the tube. Forward trailing of the band means the gradient is overloaded. Remove the band with a 1 ml syringe and 20 gauge needle. Add linear polyacrylamide and precipitate the plasmid by adding 3 volumes of EtOH. Resuspend in 50 ul of TE. Set up ligations using a constant amount of vector and increasing amounts of cDNAs. On the basis of these trial ligations, set up large scale ligation, which can be accomplished by known methods. Usually the entire cDNA prep requires 1-2 ug of cut vector.

[0061] Adaptors may be prepared by known methods, but it is preferred to resuspend crude adaptors at a concentration of 1 ug/ul, add MgSO₄ to 10 mM, and precipitate by adding 5 volumes of EtOH. Rinse with 70% EtOH and resuspend in TE at a concentration of 1 ug/ul. To kinase take 25ul of resuspended adaptors, add 3ul of 10X kinasing buffer and 20 units of kinase; incubate 37°C overnight.

[0062] Preparation of buffers mentioned in the above description of preferred methods according to the present invention will be evident to those of skill. For conve­nience, preferred buffer compositions are as follows:

Loading Buffer:	0.5 M LiCl, 10mM Tris pH 7.5, 1mM EDTA 0.1% SDS.
Middle Wash Buffer	0.15 M LiCl, 10mM Tris pH 7.5, 1mM EDTA 0.1% SDS.
Rt1 Buffer:	0.25 M Tris pH 8.8 (8.2 at 42°), 0.25 M KCl, 30 mM MgCl₂.
RT2 Buffer:	0.1 M Tris pH 7.5, 25 mM MgCl₂, 0.5 M KCl, 0.25 mg/ml BSA, 50 mM DTT.
10X Low Salt	60 mM Tris pH 7.5, 60 mM MgCl₂, 50 mM NaCl, 2.5 mg/ml BSA 70 mM Me.
10X Ligation Additions:	1mM ATP, 20 mM DTT, 1 mg/ml BSA 10 mM spermidine.
10X Kinasing Buffer:	0.5 M Tris pH 7.5, 10mM ATP, 20mM DTT, 10 mM spermidine, 1 mg/ml BSA 100 mM MgCl₂.

[0063] By "vector" is meant a DNA molecule, derived from a plasmid or bacteriophage, into which fragments of DNA may be inserted or cloned. A vector will contain one or more unique restriction sites, and may be capable of autono­mous replication in a defined host or vehicle organism such that the cloned sequence is reproducible. Thus, by "DNA expression vector" is meant any autonomous element capable of replicating in a host independently of the host's chromosome, after additional sequences of DNA have been incorporated into the autonomous element's genome. Such DNA expression vectors include bacterial plasmids and phages.

[0064] Preferred for the purposes of the present invention, however, are viral vectors, such as those derived from simian virus strain 40 (SV40). SV40 is a papovavirus having a molecular weight of 28 Mdal, and containing a circular double-stranded DNA molecule having a molecular weight of 3 Mdal, which comprises the entire genome of the virus. The entire nucleotide sequence of this single, small, covalently closed circular DNA molecule has been determined. Fiers et al., Nature 273:113-120 (1978); Reddy et al., Science 200:494-502 (1978). The viral DNA of SV40 may be obtained in large quantities, and the genomic regions responsible for various viral functions have been accurately located with respect to a detailed physical map of the DNA. Fiers et al., supra; Reddy et al., supra. The viral genome of SV40 can multiply vegetatively or as an integral part of cellular chromosomes, and a wealth of information exists on the replication and expression of this genome.

[0065] Also preferred for the purposes of the present invention is a single-stranded bacteriophage cloning vehicle, designated M13, having a closed circular DNA genome of approximately 6.5 kb. An advantage of utiliz­ing M13 as a cloning vehicle is that the phage particles released from infected cells contain single-stranded DNA homologous to only one of the two complementary strands of the cloned DNA, which therefore can be used as a template for DNA sequencing analysis.

[0066] Even more preferred for the purposes of the present invention are the expression vectors designated piH3, piH3M, and CDM8, deposited at the American Type Culture Collection (ATCC), 12301 Parklawn Drive, Rockville, MD 20852, on February 24, 1988, in E. coli piH3 has accession number ATCC 67,634, piH3M has accession number ATCC 67,633 and CDM8 has accession number ATCC 67,635.

[0067] By "tissue culture" is meant the maintenance or growth of animal tissue cells in vitro so as to allow further differentiation and preservation of cell archi­tecture or function or both. "Primary tissue cells" are those taken directly from a population consisting of cells of the same kind performing the same function in an organism. Treating such tissue cells with the proteolytic enzyme trypsin, for example, dissociates them into individual primary tissue cells that grow well seeded onto culture plates at high densities. Cell cultures arising from multiplication of primary cells in tissue culture are called "secondary cell cultures." Most secondary cells divide a finite number of times and then die. A few secondary cells, however, may pass through this "crisis period", after which they are able to multiply indefinitely to form a continuous "cell line." Cell lines often will contain extra chromosomes, and usually are abnormal in other respects as well. The immortality of these cells is a feature shared in common with cancer cells.

[0068] Preferred cell lines for use as tissue culture cells according to the present invention include the monkey kidney cell line, designated "COS." COS cells are those that have been transformed by SV40 DNA containing a functional early gene region but a defective origin of viral DNA replication. COS cell clone M6 is particularly preferred for use according to the method of the in­vention. Also preferred for the purposes of the present invention are murine "WOP" cells, which are NIH 3T3 cells transfected with polyoma origin deletion DNA. cDNA may be introduced into the host tissue culture cells of the present invention by any methods known to those of skill. Transfection may be accomplished by, for example, protoplast fusion by spheroplast fusion, or by the DEAE dextran method (Sussman et al., Cell. Biol. 4:1641-1643 (1984)).

[0069] If spheroplast fusion is employed, a preferred method is the following variant based on Sandri-Goldrin et al., Mol. Cell Bio. 1:743-752 (1981). Briefly, for example, a set of six fusions requires 100 ml of cells in broth. Grow cells containing amplifiable plasmid to OD600=0.5 in LB. Add spectinomycin to 100 ug/ml (or chloramphenicol to 150 ug/ml). Continue incubation at 37°C with shaking for 10-16 hours. (Cells begin to lyse with prolonged incubation in spectinomycin or chloramphenicol medium). Spin down 100 ml of culture (JA14/GSA rotor, 250ml bottle) 5 min. at 10,000 rpm. Drain well, resuspend pellet in bottle with 5ml cold 20% sucrose, 50mM Tris-HCL pH 8.0. Incubate on ice 5 min. Add 2 ml cold 0.25M EDTA pH 8.0, incubate 5 min. at 37°C (waterbath). Place on ice, check percent conversion to spheroplasts by microscopy. In flow hood, slowly add 20ml of cold DME/10% sucrose/10mM MgCl₂ (dropwise, ca. 2 drops per second). Remove media from cells plated the day before in 6cm dishes (50% confluent). Add 5ml of spheroplast suspension to each dish. Place dishes on top of tube carriers in swinging bucket centrifuge. Up to 6 dishes can be comfortably prepared at once. Dishes can be stacked on top of each other, but 3 in a stack is not advisable as the spheroplast layer on the top dish is often torn or detached after centrifugation. Spin at 1000xg 10 min. Force is calculated on the basis of the radius to the bottom plate. Aspirate fluid from dishes carefully. Pipet 1.5-2ml 50% (w/w) PEG1450 (or PEG1000)/50% DME (no serum) into the center of the dish. If necessary, sweep the pipet tip around to ensure that the PEG spreads evenly and radially across the whole dish. After PEG has been added to the last dish, prop all of the dishes up on their lids so that the PEG solution collects at the bottom. Aspirate the PEG. The thin layer of PEG that remains on the cells is sufficient to promote fusion; the layer remaining is easier to wash off, and better cell viability can be obtained, than if the bulk of the PEG is left behind. After 90 to 120 seconds (PEG 1000) or 120 to 150 seconds (PEG 1450) of contact with the PEG solution, pipet 1.5ml of DME (no serum) into the center of the dish. The PEG layer will be swept radially by the DME. Tilt the dishes and aspirate. Repeat the DME wash. Add 3ml of DME/10% serum containing 15 ug/ml gentamicin sulfate. Incubate 4-6 hours in incubator. Remove media and remaining bacterial suspension, add more media and incubate 2-3 days. Extensive washing of the cell layer to remove PEG tends to remove many of the cells without any substantial benefit. If the cells are allowed to sit in the second DME wash for a few minutes, most of the spheroplast layer will come up spontaneously; however it is preferred to wash briefly and allow the layer to come off in the complete medium at 37°C.

[0070] The PEG solution can be conveniently prepared by melting a fresh bottle of PEG at 60°C and pouring approx­imate 50 ml aliquots by means of a 50 ml centrifuge tube into preweighed bottles. The aliquoted PEG is stored at 5°C in the dark. To make up a fresh bottle, weigh the aliquot, remelt, and add an equal volume of DME (no serum). Adjust the pH with 7.5% Na Bicarbonate solution if necessary, and filter sterilize. The resulting PEG solution may be stored up to 3 months at room temperature without detectable adverse consequence.

[0071] Transfected host cells will be cultured according to the invention in order to accomplish expression of the protein encoded by the cDNA clone, and to increase the absolute numbers of cells available for subsequent immunoselection. Those skilled in the art will know of appropriate methods and media for this purpose, taking into account the cell type and other variables routinely considered. COS cells, for example, may be cultured in Dulbecco's modified Eagle's medium (DME) supplemented with 10% calf serum and gentamycin sulfate. Transient expression of transfected cells normally can be expected between 48 and 72 hours posttransfection. However, this time period may vary depending upon the type or strain of host cell used and the cell culture conditions, as will be apparent to those of ordinary skill.

[0072] Immunoprecipitation, blotting, and cDNA sequencing of genes cloned according to the methods of the present invention may be carried out by any convenient methods known to those of skill. For example, the immunoprecipitation protocol of Clark et al., Leukocyte Typing II, Vol. II, pp. 155-167 (1986), is preferred. Southern, Northern, or other blot analysis methods known to those of skill may be employed, using hybridization probes prepared by known methods, such as that of Hu et al. (Gene 18:271-277 (1982)). cDNA sequencing also may be accomplished by known methods, including the dideoxynucleotide method of Sanger et al., P.N.A.S. (USA) 74:5463-5467 (1977).

[0073] The antibodies used according to the present in­vention may be polyclonal or monoclonal. These may be used singly, or in conjunction with other polyclonal or monoclonal antibodies to effect immunoselection of cells expressing the desired antigen or antigens by the methods of the present invention. Methods of preparing anti­bodies or fragments thereof for use according to the present invention are known to those of skill.

[0074] Standard reference works setting forth general principles of immunology include Klein, J., Immunology: The Science of Self-Nonself Discrimination, John Wiley & Sons, publisher, New York (1982); Kennett, R., et al, eds., Laboratory Techniques in Biochemistry and Molecular Biology, Vol. 13, Elsevere, publisher, Amsterdam (1984).

[0075] The term "antibody" is meant to include the intact molecule as well as fragments thereof, such as, for example, Fab and F(ab)′₂ fragments, which also are able to bind to antigen. Polyclonal antibody preparations may be derived directly from the blood of the desired animal species after immunization with the antigen of interest, or a fragment thereof, using any of the standard proto­cols known to those of ordinary skill. Similarly, monoclonal antibodies may be prepared using known methods (Kohler et al., Eur. J. Immunol 6:292 (1976)). Use of monoclonal antibodies is preferred for the purposes of the present invention.

[0076] For the purposes of immunoselection according to the present invention, the tissue culture host cells which have been exposed to antibodies directed against the target cell surface antigen are separated from host cells which do not express the target antigen by distributing the cells onto a substrate coated with antibody directed against the antibody for the antigen. This technique, termed "panning," will be known to those of skill, and is described, for example, by Mage et al., J. Immunol. Meth. 15:47-56 (1977), and Wysocki and Sato, P.N.A.S. (USA) 75:2844-2848 (1978).

[0077] Panning according to the methods of the present invention may be carried out as follows:

a. Antibody-coated dishes. Bacteriological 60mm plates, Falcon 1007 or equivalent, or 10cm dishes such as Fisher 8-757-12 may be used. Sheep anti-mouse affinity purified antibody (from, for example, Cooper BioMedical (Cappell)) is diluted to 10ug/ml in 50mM Tris HCl, pH 9.5. Add 3ml per 6cm dish, or 10ml per 10cm dish. Let sit ca. 1.5 hrs., remove to next dish 1.5 hrs., then to 3rd dish. Wash plates 3x with 0.15 NaCl (a wash bottle is convenient for this), incubate with 3ml 1mg/ml BSA in PBS overnight, aspirate and freeze.

b. Panning. Cells will be in 60mm dishes. Aspirate medium from dish, add 2ml PBS/0.5mM EDTA/0.02% azide and incubate dishes at 37°C for 30 min. to detach cells from dish. Triturate cells vigorously with short pasteur pipet, and collect cells from each dish in a centrifuge tube. Spin 4 min. setting 2.5 (200 x g) (takes 5 min). Resuspend cells in 0.5 -1.) ml PBS/EDTA/azide/5% FBS and add antibodies. Incubate at least 30 min. on ice. Add an equal volume of PBS/EDTA/azide, layer carefully on 3 ml PBS/EDTA/azide/2% Ficoll, and spin 4 min. at setting 2.5. Aspirate supernatant in one smooth movement. Take up cells in 0.5ml PBS/EDTA/azide and add aliquots to antibody-coated dishes containing 3ml PBS/EDTA/azida/5% FBS by pipetting through 100 micron Nylon mesh (Tetko). Add cells from at most two 60mm dishes to one 60mm antibody-coated plate. Let sit at room temperature 1-3 hours. Remove excess cells not adhering to dish by gentle washing with PBS/5% serum or with medium. 2 or 3 washes of 3ml are usually sufficient.

c. Hirt Supernatant. A preferred variant of the method of Hirt, J. Molec. Biol. 26:365-369 (1967), is as follows: Add 0.4 ml 0.6% SDS, 10mM EDTA to panned plate. Let sit 20 minutes (can be as little as 1 min. if there are practically no cells on the plate). Pipet viscous mixture into microfuge tube. Add 0.1ml 5M NaCl, mix, put on ice at least 5 hrs. Keeping the mixture as cold as possible seems to improve the quality of the Hirt. Spin 4 min., remove supernatant carefully, phenol extract (twice if the first interface is not clean), add 10ug linear polyacrylamide (or other carrier), fill tube to top with EtOH, precipitate, and resuspend in 0.1ml. Add 3 volumes EtOH/NaOAc, reprecipitate and resuspend in 0.1 ml. Transform into MC1061/p3, preferably using the high efficiency protocol hereinafter described. If the DNA volume exceeds 2% of the competent cell aliquot, the transformation efficiency will suffer. 5% gives the same number of colonies as 2.5% (efficiency is halved).

[0078] It is preferred for this aspect of the present invention to use "blockers" in the incubation medium. Blockers assure that non-specific proteins, proteases, or antibodies present do not cross-link with or destroy the antibodies present on the substrate or on the host cell surface, to yield false positive or false negative results. Selection of blockers can substantially improve the specificity of the immunoselection step of the present invention. A number of non-specific monoclonal antibodies, for example, of the same class or subclass (isotype) as those used in the immunoselection step (e.g., IgG₁, IgG₂A, IgGm, etc.) can be used as blockers. Blocker concentration (normally 1-100ug/ul) is important to maintain the proper sensitivity yet inhibit unwanted interference. Those of skill also will recognize that the buffer system used for incubation may be selected to optimize blocking action and decrease non-specific binding.

[0079] A population of cells to be panned for those ex­pressing the target cell surface antigen is first de­tached from its cell culture dish (harvested) without trypsin. The cells then are exposed to a first antibody, which may be polyclonal or monoclonal, directed against the antigen of interest or against a family of related antigens. At this initial stage, a single antibody or a group of antibodies may be used, the choice depending upon the nature of the target antigen, its anticipated frequency, and other variables that will be apparent to those of skill. Target antigens expressed on the sur­faces of host cells will form an antigen-antibody com­plex.

[0080] The cells subsequently are placed in close apposition to a substrate, such as a culture dish, filter disc, or the like, which previously has been coated with a second antibody or group of antibodies. This second antibody will be directed against the first antibody, and its choice will be a matter of ordinary skill dictated by, for example, the animal in which the first antibody was raised. For example, if the first antibody was raised in mice, the second antibody might be directed against mouse immunoglobulins, raised in goats or sheep. Cells expressing the target antigen will adhere to the substrate via the complex formed between the antigen, the first antibody, and the second antibody. Adherent cells then may be separated from nonadherent cells by washing. DNA encoding the target antigen is prepared from adherent cells by known methods, such as that of Hirt, J. Molec. Biol. 26:365-369 (1967). This DNA may be transformed into E. coli or other suitable host cells for further rounds of fusion and selection, to achieve the desired degree of enrichment.

[0081] In the usual case, the initial rounds of immunoselection will employ a panel of first antibodies directed against an epitope or group of epitopes common to the family of antigens to which the target antigen belongs. This will be sufficient to narrow the number of clones for future rounds quite significantly. Two such rounds usually will be found adequate, but the number of rounds may vary as mentioned above. Thereafter, a single round of selection may be performed employing a single first antibody or a group of first antibodies recognizing only the target antigen.

[0082] By "substrate" is meant a solid surface to which antibodies may be bound for immunoselection according to the present invention. Known suitable substrates include glass, polystyrene, polypropylene, dextran, nylon, and other materials. Tubes, beads, microtiter plates, bacteriological culture dishes, and the like formed from or coated with such materials may be used. Antibodies may be covalently or physically bound to the substrate by known techniques, such as covalent bonding via an amide or ester linkage, or by absorption. Those skilled in the art will know many other suitable substrates and methods for immobilizing antibodies thereupon, or will be able to ascertain such substrates and methods using no more than routine experimentation.

[0083] The choice of host tissue culture cells for use according to the present invention preferably should be such as to avoid the situation in which the antibodies used for panning recognize determinants on untransfected cells. Thus, while COS cells are preferred for transient expression of certain surface antigens, more preferred are murine WOP cells. Of the latter, WOP 3027 cells are even more preferred. WOP cells allow virtually all antibodies to be used, since cross-reactions between murine antibodies and murine cell surface determinants are rare.

[0084] The insert size of the recombinant DNA molecule should be chosen to maximize the likelihood of obtaining an entire coding sequence. Those of skill will know various methods by which a preliminary determination of optimal insert size for a given gene may be determined.

Vector construction and cDNA insertion

[0085] Vectors suitable for expression of cDNA in mammalian tissue culture cells may be constructed by known methods. Preferred for the purposes of the present invention is an expression vector containing the SV40 origin. The vector may contain a naturally derived or synthetic tran­scription origin, and the SV40 early region promoter. Even more preferred is a chimeric promoter composed of human cytomegalovirus immediate early enhancer sequences. Various "enhancer sequences" also may be used with SV40 vectors. These are described, for example, by Banerji et al., Cell 27:299-308 (1981); Levinson et al., Nature 295:568-572 (1982); and Conrad et al., Mol. Cell. Biol. 2:949-965 (1982).

[0086] Insertion of cDNA into the vectors of the present invention can occur, for example, by homopolymeric tailing with terminal transferase. However, homopolymeric tracts located 5′ to cDNA inserts may inhibit in vitro and in vivo expression. Thus, preferred for purposes of the present invention is the use of inverted identical cleavage sites separated by a short replaceable DNA segment. Such inverted identical cleav­age sites, preferably employing the BstXI restriction endonuclease, may be used in parallel with cDNA synthetic oligonucleotides, giving the same terminii as the re­placeable segment of the vector. In this manner, the cDNA cannot ligate to itself, but can ligate to the vector. This allows the most efficient use of both cDNA and vector.

[0087] Another embodiment of the present invention is the above-described efficient oligonucleotide-based strategy to promote cDNA insertion into the vector. The piH3M vector of the present invention is preferred, and employs the inverted endonuclease sites. This vector may contain an SV40 origin of replication, but a more preferred form contains an M13 origin. This vector, containing the M13 origin, allows high level expression in COS cells of coding sequences placed under its control. Also, the small size and particular arrangement of sequences in the plasmid permit high level replication in COS cells.

[0088] By "cell surface antigen" is meant any protein that is transported through the intracellular membrane system to the cell surface. Such antigens normally are anchored to the cell surface membrane through a carboxyl terminal domain containing hydrophobic amino acids that lie in the lipid bilayer of the membrane, and there exert their biological and antigenic effects. Antigens such as those of T-lymphocytes are particularly suited for gene cloning by the method of the present invention. However, cell surface antigens of any cells may be cloned according to the present method. Moreover, proteins not normally expressed on the cell surface may admit of cloning according to the present method by, for example, using fluorescence activated cell sorting (FACS) to enrich for fixed cells expressing intracellular antigens.

[0089] By "substantially pure" is meant any antigen of the present invention, or any gene encoding any such antigen, which is essentially free of other antigens or genes, respectively, or of other contaminants with which it might normally be found in nature, and as such exists in a form not found in nature. By "functional derivative" is meant the "fragments," "variants," "analogs," or "chemical derivatives" of a molecule. A "fragment" of a molecule, such as any of the antigens of the present invention, is meant to refer to any polypeptide subset of the molecule. A "variant" of such molecules is meant to refer to a naturally occurring molecule substantially similar to either the entire molecule, or a fragment thereof. An "analog" of a molecule is meant to refer to a non-natural molecule substantially similar to either the entire molecule or a fragment thereof.

[0090] A molecule is said to be "substantially similar" to another molecule if the sequence of amino acids in both molecules is substantially the same, and if both mole­cules possess a similar biological activity. Thus, provided that two molecules possess a similar activity, they are considered variants as that term is used herein even if one of the molecules contains additional amino acid residues not found in the other, or if the sequence of amino acid residues is not identical. As used herein, a molecule is said to be a "chemical derivative" of another molecule when it contains additional chemical moieties not normally a part of the molecule. Such moieties may improve the molecule"s solubility, absorp­tion, biological half life, etc. The moieties may alternatively decrease the toxicity of the molecule, eliminate or attenuate any undesirable side effects of the molecule, etc. Moieties capable of mediating such effects are disclosed, for example, in Remington's Pharmaceutical Sciences, 16th ed., Mack Publishing Co., Easton, Penn. (1980).

[0091] Similarly, a "functional derivative" of a gene of any of the antigens of the present invention is meant to include "fragments," "variants," or "analogues" of the gene, which may be "substantially similar" in nucleotide sequence, and which encode a molecule possessing similar activity.

[0092] The substantially pure antigens that have been expressed by methods of the present invention may be used in immunodiagnostic assay methods well known to those of skill, including radio-immunoassays (RIAs), enzyme immunoassays (EIAs) and enzyme-linked immunosorbent assays (ELISAs). The substantially pure proteins of the present invention, in soluble form, may be administered alone or in combination with other antigens of the present invention, or with other agents, including lymphokines and monokines or drugs, for the treatment of immune-related diseases and disorders in animals, including humans. As examples of such disorders that may benefit from treatment with the substantially pure proteins of the present invention may be mentioned immune deficiency diseases, diseases of immediate type hypersensitivity, asthma, hypersensitivity pneumonitis, immune-complex disease, vasculitis, systemic lupus erythematosus, rheumatoid arthritis, immunopathogenic renal injury, acute and chronic inflammation, hemolytic anemias, platelet disorders, plasma and other cell neoplasms, amyloidosis, parasitic diseases, multiple sclerosis, Guillain-Barre syndrome, acute and subacute myopathic paralysis, myasthenia gravis, immune endocrinopathies, and tissue and organ transplant re­jection, all as described in Petersdorf et al., eds., Harrison's Principles of Internal Medicine, supra. See also Weir, ed., supra; Boguslaski et al., eds., supra; and Holborow et al., eds., supra.

[0093] When used for immunotherapy, the antigens of the present invention may be unlabeled or labeled with a therapeutic agent. Examples of therapeutic agents which can be coupled to the antigens of the invention for immunotherapy are drugs, radioisotopes, lectins, and toxins.

[0094] The dose ranges for the administration of the antigens of the present invention are those large enough to produce the desired immunotherapeutic effect, but not so large as to cause adverse side effects, such as unwanted cross-reactions, anaphylactic reactions, and the like. Generally, the dosage employed will vary with the age, condition, sex, and extent of the disease in the patient. Counterindications (if any), immune tolerance and other variables also will affect the proper dosage. Administration may be parenteral, by injection or by gradual perfusion over time. Administration also may be intravenous, intraparenteral, intramuscular, subcutaneous, or intradermal.

[0095] Preparations for parenteral administration include sterile or aqueous or non-aqueous solutions, suspensions and emulsions. Examples of non-aqueous solvents include propylene glycol, polyethylene glycol, vegetable oils such as olive oil, and injectable organic esters such as ethyl oleate. Aqueous carriers include water, alcoholic and aqueous solutions, emulsions, or suspensions, including saline and buffered media. Parenteral vehicles include sodium chloride solution, Ringer's dextrose, dextrose and sodium chloride, lactated Ringer's, or fixed oils. Intravenous vehicles include fluid and nutrient replenishers, electrolyte replenishers, such as those based on Ringer's dextrose, and the like. Preservatives and other additives also may be present, such as, for example, antimicrobials, antioxidants, chelating agents, inert gases and the like. Such preparations, and the manner and method of making them, are known and de­scribed, for example, in Remington's Pharmaceutical Science, 16th ed., supra.

[0096] The antigens of the present invention also may be prepared as medicaments or pharmaceutical compositions comprising the antigens, either alone or in combination with other antigens or other agents such as lymphokines, monokines, and drugs, the medicaments being used for therapy of animal, including human, immune-related indications.

[0097] Although the antigens of the present invention may be administered alone, it is preferred that they be administered as a pharmaceutical composition. The compositions of the present invention comprise at least one antigen or its pharmaceutically acceptable salt, together with one or more acceptable carriers and option­ally other therapeutic agents. By "acceptable" is meant that the agent or carrier be compatible with other ingredients of the composition and not injurious to the patient. Compositions include those suitable for oral, rectal, nasal, topical (including buccal and sublingual), vaginal, or parenteral administration. The compositions conveniently may be presented in unit dosage form, and may be prepared by methods well known in the pharmaceuti­cal arts. Such methods include bringing into association the active ingredients with the carrier which constitutes one or more accessory ingredients. In general, composi­tions are prepared by uniformly and intimately bringing into association the active ingredient with liquid carriers or finely divided solid carriers, or both, and shaping the product formed thereby, if required.

[0098] Orally administered pharmaceutical compositions according to the present invention may be in any conve­nient form, including capsules, cachets, or tablets, each containing a predetermined amount of the active ingredi­ent. Powders or granules also are possible, as well as solution or suspension in aqueous or nonaqueous liquids, or oil-in-water liquid emulsions, or water-in-oil liquid emulsions. The active ingredient also may be presented as a bolus, electuary or paste.

[0099] Having now described the invention, the same will be more fully understood by reference to the following examples, which are not intended in any way to limit the scope of the invention.

EXAMPLE I Isolation, Molecular Cloning, and Structure of the Human CD2 Antigen

The cDNA expression vector piH3

[0100] A COS cell expression vector was constructed from piSV (Little et al., Mol. Biol. Med. 1:473-488 (1983)) by inserting a synthetic transcription unit between the suppressor tRNA gene and the SV40 origin. The tran­scription unit consisted of a chimeric promoter composed of human cytomegalovirus AD169 immediately early enhancer sequences fused to the HIV LTR -67 to +80 sequences. Immediately downstream from the LTR +80 sequence was inserted a polylinker containing two BstXI sites separat­ed by a 350bp stuffer; the BstXI sites were flanked by Xbal sites, which could also be used to excise the insert. Downstream from the polylinker were placed the SV40 small t antigen splice and early region polyadenylation signals derived from pSV2. The nucleotide sequence of the vector is shown in Figure 1.

cDNA library construction

[0101] RNA was prepared from HPB-ALL cells by the guanidinium thiocyanate/CsCl method, as described above. PolyA⁺ RNA was prepared from total RNA by oligo dT selection. Maniatis et al, Molecular Cloning: A Labo­ratory Manual, supra. cDNA was synthesized by the method of Gubler and Hoffman, (Gene 25:263-269 (1982)). BstXI adaptors were ligated to the cDNA, and the reaction products fractionated by centrifugation through a 5ml-20% potassium acetate gradient containing 1mM EDTA for 3 hours at 50krpm in a SW55 rotor. 0.5ml fractions were collected manually through a syringe needle or butterfly inserted just above the curve of the tube. Individual fractions were ethanol-precipitated after addition of linear polyacrylamide (Strauss and Varshavsky, Cell 37:889-901 (1984)) to 20 ug/ml. Fractions containing cDNA larger than 700bp were pooled and ligated to gradi­ent purified BstXI digested piH3 vector.

[0102] The ligated DNA was transformed into E. coli MC1061/p3 made competent by the following protocol: The desired strain was streaked out on an LB plate. The next day a single colony was inoculated into 20ml TYM broth (recipes below) in a 250ml flask. The cells were grown to midlog phase (OD₆₀₀ about 0.2-0.8), poured into a 2l flask containing 100ml TYM, and vigorously agitated until cells grew to 0.5-0/9 OD, then diluted again to 500ml in the same vessel. When the cells grew to OD₆₀₀ 0.6, the flask was placed in ice-water, and shaken gently to assure rapid cooling. When the culture was cool, it was spun at 4.2k rpm for 15 minutes (J6). The supernatant was poured off and the pellet resuspended in about 100ml cold TfB I (below) by gentle shaking on ice. Thereafter, it was respun in the same bottle at 4.2krpm for 8 minutes (J6). The supernatant was poured off and the pellet resuspended in 20 ml cold TfB II by gentle shaking on ice. 0.1 to 0.5 ml aliquots were placed in prechilled microfuge tubes, frozen in liquid nitrogen, and stored at -70°C. For transformation, an aliquot was removed, thawed at room temperature until just melting, and placed on ice. DNA was added, let sit on ice 15-30 minutes, and incubated at 37°C for 5 minutes (6 minutes for 0.5ml aliquots). Thereafter the DNA-containing suspensions were diluted 1:10 in LB and grown for 90 minutes before plating or applying antibiotic selection. Alternatively, the heat-pulsed transformation mix was plated directly on antibiotic plates onto which a thin (4-5ml) layer of antibiotic-free LB agar was poured just before plating.

[0103] Media and Buffers: TYM: 2% Bacto-Tryptone, 0.5% Yeast Extract, 0.1M NaCl, 10mM MgsO₄ (can be added before autoclaving). TfB I: 30mM KOAc, 50mM MnCl₂, 100mM KCL, 10mM CaCl₂, 15% (v/v) glycerol. TfB II: 10mM Na-MOPS, pH 7.0, 75 MM CaCl₂, 10mM KCl, 15% glycerol.

Recovery of cDNA clones by panning

[0104] Bacteriological culture dishes (Falcon 1007) were prepared for panning by coating with an affinity purified sheep anti-mouse IgG antibody as described by Wysocki and Sato (Proc. Natl. Acad. Sci. USA 75:2844-2848 (1978)), except that dishes were washed with 0.15M NaCl from a wash bottle instead of PBS, and unreacted sites were blocked by overnight incubation in PBS containing 1mg/ml BSA. Dishes were typically prepared in large batches and stored frozen, after aspiration of the PBS/BSA. In the first round of screening, 24 6cm dishes of 50% confluent COS cells were transfected by protoplast fusion according to the method of Sandri-Goldrin et al., Mol. Cell. Biol. 1:743-752 (1981). 72 hours post fusion the cells were detached by incubation in PBS/1mM EDTA/.02% sodium azide at 37°C for 30 minutes. The detached cells were pooled, centrifuged, and resuspended in cold PBS/EDTA/5% Fetal Bovine Serum containing monoclonal antibodies, usually as ascites at 1:1000 dilution, but also as commercial reagents at the concentrations suggested by the manufac­tures. After 1 hour on ice, the cells were diluted with 1:1 with PBS/EDTA/azide and layered on 10ml of PBS/EDTA/­azide containing 2% Ficoll 400. After centrifugation (400xg, 5 minutes), the supernatant was carefully as­pirated, the pellet resuspended in a small amount of PBS/EDTA/5% FBS, and the cells distributed into panning plates containing 3 ml of PBS/EDTA/5% FBS. The plates were then treated essentially as described by Wysocki and Sato. Proc. Natl. Acad. Sci. USA 75:2844-2848 (1978). Episomal DNA was recovered from the adherent cells by the Hirt (J. Mol. Biol. 26:365-269 (1967)) procedure and transformed into MC1061/p3.

Cell lines and cell culture

[0105] COS cell clone M6 cells were propagated in Dulbecco's modified Eagle's medium supplemented with 10% calf serum and gentamycin sulfate at 15ug/ml (DME/10% calf serum). Cells were split the day before transfection in 6 cm dishes at approximately 1:8 ratio from stock plates kept as dense as possible without overtly affronting the cells. T cell lines were grown in Iscove's modification of Dulbecco's medium (IMDM) con­taining gentamycin as above, and either NuSerum (Collab­orative Research) or fetal bovine serum at 10%.

COS cell transfection for immunofluorescence studies

[0106] COS cells at 50% confluence in 6 cm dishes were transfected in a volume of 1.5ml with a cocktail consist­ing of DME or IMDM medium containing 10% NuSerum (Collab­orative Research), 400ug/ml DEAE Dextran, 10uM chloroquine diphosphate, and 1ug/mg DNA. After 4 hours at 37°C (or earlier if the cells appeared ill), the transfection mix was removed and the cells were treated with 10% DMSO in PBS for 2 minutes. Sussman and Milman, Cell Biol. 4:1641-1643 (1984). Cells were then returned to DME/10% calf serum for 48 to 72 hours to allow ex­pression.

Immunoprecipitations, Northerns and Southerns

[0107] T cells were labeled by lactoperoxidase treatment, lysed, and immunoprecipitated by the procedure of Clark and Einfeld (Leukocyte Typing II, Vol. II, pp. 155-167 (1986)), using commercially available goat anti-mouse IgG agarose beads (Cooper Biomedical). COS cells were transfected with DEAE Dextran method and trypsinized and passed without dilution into new plates 24 hours after transfection. 36 hours later, cells were detached by exposure to PBS/EDTA as above, centrifuged, and labeled by the lactoperoxidate method. A cleared lysate was prepared as for the T cell immunoprecipitations, except that the lysis buffer contained 1mM PMSF, and incubation with the primary antibody was carried out for only 2 hours at 4°C. Eluted samples were fractionated on discontinuous 11.25% polyacrylamide gels using the buffer system of Laemmli (Nature 227:680-685 (1970)).

[0108] Northern blot analysis was carried out essentially as described (Maniatis et al., Molecular Cloning, a Laboratory Manual (1982)), except that DMSO was omitted from the loading buffer, denaturation was at 70°C for 5 minutes, and the gel contained 0.6% formaldehyde rather than 6%. The gel was stained in two volumes of water containing 1ug/ml ethidium bromide, photographed, and transferred to nylon (GeneScreen, DuPont) in the staining liquor. The transferred RNA was irradiated by exposure to a germicidal lamp through Saran Wrap (Church and Gilbert, Proc. Natl. Acad. Sci. USA 8:1991-1995 (1984)) for 5 minutes at a flux (measured at 254nm) of 0.22mW/cm². Southern blot analysis was carried out by alkaline transfer to nylon (GeneScreen, DuPont) as described by Reed and Mann (Nucl. Acids Res. 13:7207-7221 (1986)). Hybridization probes were prepared by the method of Hu and Messing (Gene 18:271-277 (1982)), and blots were prehybridized in SDS/phosphate buffer (Church and Gilbert, Proc. Natl. Acad. Sci. USA 8:1991-1995 (1984)) containing 10 DNA microgram equivalents of M13 mp19 phage.

Erythrocyte Rosetting

[0109] Erythrocytes were prepared from whole blood by three centrifugations in PBS. COS cells were transfected in 6 cm dishes with CD2 or other surface antigen expression clones by the DEAE method. 48 to 72 hours posttransfection, the medium was aspirated and 2ml of PBS/5% FDS/azide was added to each plate, followed by 0.4 ml of the appropriate erythrocyte samples as 20% suspensions in PBS. After 1 hour at room temperature, the nonadherent erythrocytes were gently washed off, and the plates examined.

[0110] A cDNA encoding CD2 antigen determinants was isolated in the following manner: cDNA was prepared from RNA extracted from the human T Cell tumor line HPB-ALL and inserted into the SV40 origin-based expression vector piH3 as described above. A cDNA library of approximately 3 x 10⁵ recombinants was constructed, and the library was introduced into COS cells by protoplast fusion. Three days later the cells were detached by exposure to EDTA and treated with a pool of monoclonal antibodies, including three (OKT11, Leu5b, and Coulter T11) directed against CD2 determinants. The antibody-treated cells were distributed into dishes coated with an affinity purified sheet anti-mouse IgG antibody, allowed to attach, and separated from the nonadherent cells by gentle washing. This method of enrichment is known in the immunological literature (Mage et al., J. Immunol. Methods 15:47-56 (1977).

[0111] The resulting colonies were pooled, fused into COS cells, and subjected to a second round of panning as before. In the third round, a portion of the detached cells was treated with a mixture of three monoclonal antibodies specific for CD2, and a Hirt supernatant was again generated and transformed into E. coli. DNA was prepared from eight of the resulting colonies and transfected into COS cells. After three days, surface expression of the CD2 antigen was detected by indirect immunofluorescent in six of eight transfected dishes. Restriction enzyme digestion of the corresponding plasmid DNAs revealed a 1.5kb insert in all six isolates.

[0112] One of the six clones was prepared in larger quan­tities for further analysis. Following transfection into COS cells, indirect immunofluorescence analysis with a partial panel of antibodies provided by the Third Inter­national Workshop on Leukocyte Differentiation Antigens showed that all of the antibodies provided gave positive reactions with the exception of one sample which also failed to react with phytohemagglutinin-activated T lymphocytes. Among the 17 antibodies tested were at least eight distinguishable groups defined by their differing patterns of reactivity with lymphocytes of various primate species. Jonker and Nooij, Leukocyte Typing II, Vol. I, pp. 373-387 (1986).

cDNA sequence analysis

[0113] The CD2 cDNA insert was subcloned into M13 mp19 (Vieira and Messing, Gene 19: 259-268 (1982)) in both orientations, and the sequence determined by the dideoxynucleotide method (Figure 2). Sanger et al., Proc. Natl. Acad. Sci. USA 74: 5464-5467 (1977). An open reading frame was observed to extend 360 residues from an ATG triplet satisfying the consensus criteria of Kozak (Microbiol. Rev.: 1-47:45 (1983)) for translational initiation codons (Figure 1). The predicted amino acid sequence evokes an integral membrane protein with a single membrane spanning hydrophobic anchor terminating in a rather large intracytoplasmic domain. Comparison of the N-terminal amino sequence with the matrix of signal sequence residue frequencies constructed by von Heijne (Nucl. Acids Res. 14:4683-4690 (1986)) suggests that mature CD2 peptide is formed by cleavage of a precursor peptide between the 19th (Ser) and 20th (Lys) residues.

[0114] A surprising and unexpected feature of this sequence is the presence of a potential N-linked glycosylation site just proximal to the proposed cleavage site. The resulting polypeptide backbone has a predicted molecular weight of 38.9kd divided into an external domain of mass 21.9kb and a cytoplasmic domain of mass 1.6kd. Three N-linked glycosylation sites are present in the extra-­cellular domain.

[0115] The membrane spanning domain comprises 26 unchanged residues of predominantly hydrophobic character. In the nine residues immediately following are seven basic residues, either lysines or arginines. The appearance of predominantly hydrophobic residues followed by basic residues is a common organizational feature of transmembrane proteins bearing carboxyl-terminal anchors.

[0116] Another surprising feature of the transmembrane domain is the appearance of a cys-gly-gly-gly, a beta turn motif (Chou and Fasman, Annual Review of Biochemistry, 47:251-276 (1978)), flanked by hydrophobic residues (which are frequently found flanking beta turns). Because only 20 residues arrayed in an alpha helix are theoretically needed to traverse the 3nm membrane bilayer (Tanford, Science, 200:1012-1018 (1978)), and as few as 14 hydrophobic residues can allow insertion and export of an integral membrane protein (Adams and Rose, Cell, 41:1007-1015 (1985)), the trans­membrane segment of the CD2 antigen may contain a bend or king.

[0117] The rather large size of the cytoplasmic domain leaves open the possibility that CD2 possesses an intrinsic enzymatic activity. The cytoplasmic domain is very rich in prolines and contains three sites with high turn probability.

[0118] Comparison of the amino acid sequence with the NBRF database revealed no substantive homologies with other proteins. In particular, no homology with the T cell receptor alpha or beta chains was observed, ruling out the suggestion that CD2 is a primordial T cell receptor. Milanese et al., Science, 231:1118-1122 (1986).

[0119] Just inside the cytoplasmic face of the protein is a run of basic proteins followed by a serine residue, a pattern found at the same location in both the EGF receptor and the class I histocompatibility genes, and in each case a known site for either in vivo (EGF) and in vitro (HLA) phosphorylation by protein kinase C or cyclic AMP-dependent protein kinase, respectively. Hunter et al., Nature, 311:480-482 (1984); David and Czech (1985); Guild and Strominger (1984). A similar site is found in the intracytoplasmic domain of the interleukin 2 receptor, and is phosphorylated in vivo by protein kinase C. Leonard et al., Nature, 311:626-631 (1984); Nikaido et al., Nature, 311:631-635 (1984); Shackelford and Trowbridge, J. Biol. Chem., 259:11706-     (1984).

Immunoprecipitation of CD2 antigen expressed by transfected cells

[0120] COS cells were transfected with the CD2 expression plasmid and surface labeled with 125_I by the lactoperoxidase method 60 hours post-transfection. A cell lysate was prepared, and portions were incubated either with monoclonal anti-CD2 antibody (OKT11) or with an extraneous (OKT4; anti-CD4) antibody for 2 hours at 4°C. Sepharose-bound anti-mouse antibody was added, and after several washing steps, the adsorbed proteins were eluted and electrophoresed through a 11.25% acrylamide gel together with similarly prepared immunoprecipitates from phytohemagglutinin-activated T lymphocytes, the cDNA donor line HPB-ALL, or a long-term T cell line generated in this laboratory. Autoradiography demonstrated a prominent band of immunoreactive material precipitated from transfected COS cells by the anti-CD2 antibody, but not by the control. The calculated mean molecular weight of the COS cell material was 51kd, compared to a mean molecular weight of 54kd for the T blast and T cell line material; the antigen from HPB-ALL cells was found to have a molecular weight of approximately 61kd. The observed differences in size were attributed to different patterns of glycosylation in the different cell types. A minor band of apparent molecular weight 38kd was present in material immunoprecipitated from COS cells but not from T cells or HPB-ALL cells. The size of this species agrees within experimental error with the predicted molecular weight of mature unglycosylated peptide, 39kd.

COS cells expressing CD2 form rosettes with sheep erythrocytes

[0121] COS cells transfected with the CD2 expression clone were treated for 1 hour with purified MT910 (IgG, kappa) anti-CD2 antibody (Rieber et al., Leukocyte Typing II, Vol. I, pp. 233-242 (1986)) at a concentration of 1ug/ml, or with purified MG40.5 (IgGl, kappa; Kawata et al., J. Exp. Med., 160:633-651 (1984)) antibody at the same concentration. MB40.5 recognizes a monomorphic HLA-ABC determinant and cross-reacts with African Green Monkey histocompatibility antigens; it was chosen because it represents an isotype-matched antibody recognizing a surface antigen of approximately the same abundance as the CD2 antigen expressed by transfected cells. Sheep erythrocyte rosettes were observed in the presence of MF40.5, but not of MT910. Rosette inhibition was also observed with OKT11 antibody, and not with various other control antibodies.

Transfected COS cells form rosettes with other animal erythrocytes

[0122] In addition to sheep erythrocytes, human T cells are known to form rosettes with horse, pig, dog, goat, and rabbit, but not mouse or rat erythrocytes. Johansen et al., J. Allergy Clin. Immunol., 54:86-94 (1974); Amiot et al. in, A. Bernard et al., eds., Leucocyte Typing, Springer, publisher, New York, N.Y., pp. 281-293 (1984); Nalet and Fournier, Cell. Immunol., 96:126-136 (1985). Autorosettes between human erythrocytes and human thymocytes (Baxley et al., Clin. Exp. Immunol., 15:385-393 (1973)) have also been reported. COS cells transfected with the CD2 expression clone were treated with either MT910 or with the control antibody, MB40.5, and exposed to erythrocytes from the species above. Rosettes were observed with horse, pig, dog, goat, sheep, rabbit, and human erythrocytes, but not with mouse or rat erythrocytes. Rosette formation was blocked by pretreatment of transfected COS cells with MT910, but not with MB40.5. In these experiments, it was noticed that horse erythrocytes formed unusually dense rosettes, and that goat erythrocytes formed rather sparse rosettes, possibly because their small size made them more susceptible to washing. Mouse erythrocytes showed weak spontaneous binding to the culture dish as well as to MT910 and MB40.5 pretreated cells, while rat erythrocytes showed no detectable binding of any sort.

Binding of human erythrocytes is blocked by LFA3 antibody

[0123] Because it has been suggested on the basis of antibody blocking studies that LFA3 is the target structure for the CD2 antigen (Shaw et al., Nature, 323:262-264 (1986)), the ability of anti-LFA3 antibody to prevent rosette formation was investigated. Transfected cells were exposed to human erythrocytes pretreated for 2 hours with either anti-LFA3 (IgGl, kappa) as ascites at 1:1000 dilution, or with a 10ug/ml concentration of each of four isotype-matched nonagglutinating antibodies directed against human erythrocyte antigens as prevalent or more prevalent than LFA3:G10/B11 and D10, anti-K14 antigen, D6, anti-Wr^b antigen; and F7/B9, anti-k antigen. Nichols et al., Vox Sang, in press. The erythrocytes were washed free of excess LFA3 antibody, but were allowed to form rosettes in the presence of the control antibodies to guard against possible loss of antibody blocking power by desorption. Rosette formation was observed in the presence of all four control antibodies, but not with erythrocytes pretreated with anti-LFA3.

COS cells expressing other T cell antigens do not form rosettes

[0124] A number of clones were isolated by the same expression technique used to clone CD2 and characterized to varying degrees by antibody reactivity, nucleic acid restriction and sequence analysis, and immunoprecipi­tation. Representative clones were transfected into COS cells and analyzed for ability to sustain rosette formation. The CD1a, CD1b, CD1c, CD4, CD5, CD6, CD6, CD8, and CD27 (Tp44) clones did not form rosettes with human erythrocytes.

RNA blot analysis

[0125] Equal amounts of total RNA prepared from cell types expressing or lacking CD2 antigen were electrophoresed through denaturing agarose gels and transferred to nylon. Hybridization of the transferred RNA with a strand selective probe (Hu and Messing, Gene, 17:271-277 (1982)) prepared from an M13 clone containing a CD2 cDNA insert revealed the presence of prominent 1.65 and 1.3 kb transcripts present in RNA derived from thymocyte, activated T cell, and senescent T cell populations. Lesser amounts were found in RNA extracted from cDNA donor line, HPB-ALL and less still from MOLT4; barely detectable levels were recorded in RNA from the HSB-2 line. No reactivity was observed with RNA from Namalwa (Burkitt lymphoma), U937 (histiocytic leukemia), HuT-78 (Adult T cell leukemia), PEER (T cell leukemia), or Jurkat clone J3R7 (T cell leukemia) lines. The pattern genomic alteration underlies the failure of the examined T cell tumor lines to express CD2 antigen. Restriction analysis of total genomic DNA with a number of other enzymes, as well as preliminary results with an incomplete collection of 1 phage recombinants bearing the CD2 sequence, shows that the gene is divided into at least four exons.

EXAMPLE II Isolation and Molecular Cloning of Human LFA-3 Antigen

[0126] The previous example shows that cDNAs encoding surface antigens, such as the CD2 antigen, can be isolated by the transient expression system of the present invention, in which COS cells transfected with cDNA libraries are allowed to attach to ("panned" on) antibody-coated plates. Plasmid DNA is recovered from cells adhering to the plates, transformed into E. coli, and the process is repeated, usually twice, to isolate the desired clone. Although powerful, this approach cannot be used when the monoclonal antibodies used for panning recognize determinants on the untransfected cells. This appears to be the case for anti-LFA3 monoclonal TS2/9. However, a similar transient expression system based on polyoma virus replication-­competent cells should allow almost all monoclonals to be used, since the probability of cross reaction between murine antibodies and murine cells surface determinants should usually be small.

[0127] A new expression vector, CDM8 (Figure 3) was created from the COS cell vector piH3M described previously. The new vector differs by the inclusion of a deleted version of a mutant polyoma virus early region selected for high efficiency expression in both murine and monkey cells, by the replacement of substantially all of the human immunodeficiency promoter region with the cognate sequences of the human cytomegalovirus immediate early promoter, and by inclusion of a bacteriophage T7 promoter between the eukaryotic promoter and the side of cDNA insertion. Expression in COS cells of chloramphenicol acetyltransferase by all of the vectors was equivalent.

[0128] A library of 1.9 x 10⁶ recombinants having inserts greater than 0.8kb in size was prepared in the CDM8 vector from a microgram of poly A+ RNA isolated from the human lymphoblastoid cell line JY. The library was introduced into WOP cells (NIH 3T3 cells transfected with polyoma origin deletion DNA) by spheroplast fusion, and subjected to three rounds of panning and reintroduction into E. coli as described in Example I.

[0129] A clone encoding the LFA-3 antigen was identified by indirect immunofluorescence of transfected WOP cells, amplified and sequenced (Figure 4). Within the 874 bp insert, an open reading frame of 237 residues originates at a methionine codon closely corresponding to the consensus sequence suggested by Kozak, Microbiol. Rev., 47:1-45 (1983). The reading frame terminates in a series of hydrophobic residues lacking the characteristic basic anchoring residues of internal membrane proteins, but sharing features with known phosphatidylinositol-linked superficial membrane proteins. The features include clustered serine or threonine residues in a hydrophilic region immediately preceding the hydrophobic domain, and the presence of serines and threonines in the hydrophobic portion.

[0130] The amino acid sequence predicted from the nucleotide sequence of the LFA-3 clone was compared to the NBRF database, and no significant homologies were uncovered; the most significant scores were to the HIV envelope protein. Within the 200 residues comprising the presumed mature protein are 6 N-linked glycosylation sites, and 5 tandem serine or tandem threonine residues that frequently appear in O-linked glycosylated proteins. Ten cysteine residues appear in the complete sequence, 6 of which are distributed in the latter half of the mature protein, and one of which falls in the carboxy-terminal hydrophobic domain. Although esterification of cysteine thiols to fatty acids is a common occurrence in integral membrane proteins, and may play an alternate role in membrane anchoring of LFA-3, two examples are known of cysteine residues within or at the margin of the hydrophobic region of phosphatidylinositol linked proteins.

[0131] The predicted sequence suggests that the known manipulations for increasing erythrocyte adhersion to T cells may find direct physical explanation in the structure of the LFA-3 molecule. Aminoethylisothiouroni­um bromide, the thiourea adduct of bromoethylamine, undergoes spontaneous rearrangement to mercaptoethyl­guanidine at alkaline pH. The latter likely gains access to disulfide bonds inaccessible to less chaotropic reducing agents and may thereby reduce and promote the unfolding of the LFA-3 molecule. Similarly, neuraminidase may decrease steric interference by the many carbohydrate chains on the molecule.

[0132] RNA and DNA blot hybridization analysis showed that the LFA-3 gene shares no closely related sequences in the genome, and encodes a single RNA species of about 1 kb in length. Cell lines that express large amounts of surface LFA-3 have greater amounts of LFA-3 RNA than those that express small or nondetectable amounts.

[0133] Radioimmunoprecipitation of the antigen expressed in transfected COS and murine cells shows a broad band of approximately 50kd mean molecular mass, similar to that found in JY cells.

EXAMPLE III Isolation and Molecular Cloning of the Human CD28 cDNA Antigen

[0134] The previous examples illustrate the monoclonal antibody-based technique of the present invention for enrichment of cDNAs encoding surface antigens. In the present example, a method of constructing plasmid expression libraries is described which allows the enrichment technique to be fully exploited. The method of the present invention for making plasmid expression libraries is of general use for expression cloning.

[0135] The antibody selection technique of the present invention has also been applied to isolate a cDNA clone encoding the CD28 antigen. The antigen shares substantial homology with members of the immunoglobulin superfamily and forms a dimer structure on the surface of transfected COS cells similar to the dimer structure found on T lymphocytes.

Preparation of cDNA Libraries.

[0136] Poly(A)+ RNA was prepared from the human T-cell tumor line HPB-ALL by oligo(dT) cellulose chromatography of total RNA isolated by the guanidinium thiocyanate method (Chirgwin, J.M. et al., Biochemistry, 18:5294-5299 (1979)). cDNA was prepared by a protocol based on the method of Gubler and Hoffman (Gubler, U. et al., Gene, 25:263-269 (1982)). 4ug of mRNA was heated to approximately 100°C in a 1.5 ml centrifuge tube for 30 seconds, quenched on ice, and the volume adjusted to 70ul with RNAse-free water. To this were added 20ul of buffer (0.25 M Tris pH 8.8 (8.2 at 42°C) 0.25 M KCl, 30 mM MgCl₂), 2ul of RNAse inhibitor (Boehringer 36 u/ul), 1 ul of 1M DTT, 1ul of 5ug/ul of oligo dT (Collaborative Research), 2ul of 25 mM each deoxynucleoside triphosphate (US Biochemicals), and 4ul of reverse transcriptase (Life Sciences, 24u/ul). After 40 minutes at 42°C, the reaction was terminated by heating to 70°C for 10 minutes. To the reaction mix was then added 320ul of RNAse free water, 80ul of buffer (0.1 M Tris pH 7.5, 25 mM MgCl₂, 0.5 M KCl, 0.25 mg/ml BSA, and 50 mM (DTT), 25 units of DNA Polymerase I (Boehringer), and 4 units of RNAse H (BRL). After 1 hour at 15°C and 1 hour at 22°C, 20 ul of 0.5M EDTA pH 8.0 were added, the reaction mixture was extracted with phenol, NaCl was added to 0.5 M, linear polyacrylamide (carrier; Strauss, F. et al., Cell, 37:889-901 (1984)) was added to 20 ug/ml, and the tube was filled with ethanol. After centrifugation for 2-3 minutes at 12,000 x g, the tube was removed, vortexed to dislodge precipitate spread on the wall of the tube, and respun for 1 minute.

[0137] Unpurified oligonucleotides having the sequence CTCTAAAG and CTTTAGAGCACA were dissolved at a concentration of 1 mg/ml, MgSO₄ was added to 10 mM, and the DNA precipitated by adding 5 volumes of EtOH. The pellet was rinsed with 70% ETOH and resuspended in TE at a concentration of 1 mg/ml. 25 ul of the resuspended oligonucleotides were phosphorylated by the addition of 3 ul of buffer (0.5 M Tris pH 7.5, 10 mM ATO, 20 mM DTT, mM spermidine, 1 mg/ml BSA, and 10 mM MgCl₂) and 20 units of polynucleotide kinase followed by incubation at 37°C overnight.

[0138] 3 ul of the 12-met and 2 ul of the 8-mer phosphorylated oligonucleotides were added to the cDNA prepared as above in a 300 ul reaction mixture containing 6 mM Tris pH 7.5, 6 mM MgCl₂, 5 mM NaCl, 0.35 mg/ml BSA, 7 mM mercaptoethanol, 0.1 mM ATP, 2 mM DTT, 1 mM spermidine and 400 units T4 DNA ligase (New England BioLabs) at 15° overnight. 10 ul of 0.5 M EDTA were added, the reaction was phenol extracted, ethanol precipitated, resuspended in a volume of 100 ul and layered on a 5 ml gradient of 5-20% potassium acetate in 1 mM EDTA, 1 ug/ml ethidium bromide. The gradient was spun 3 hours at 50,000 rpm (SW55 rotor) and fractionated manually, collecting three approximately 0.5 ml fractions followed by six approximately 0.25 fractions in micro­centrifuge tubes by means of a butterfly infusion set inserted just above the curve of the tube. Linear polyacrylamide was added to 20 ug/ml, the tubes were filled with ethanol, chilled, spun, vortexed and respun as above. The precipitate was washed with 70% ethanol, dried, and resuspended in 10 ul. 1 ul of the last 6 fractions was run on a gel to determine which fractions to pool, and material less than 1kb in size was typically discarded. Remaining fractions were pooled and ligated to the vector.

[0139] The complete sequence and derivation of the vector is shown in Figure 5. The vector was prepared for cloning by digestion with BstXI and fractionation on 5-20% potassium acetate gradients as described for the cDNA. The appropriate band was collected by syringe under 300 nm UV light and ethanol precipitated as above. cDNA and vector were titrated in test ligations. Usually 1-2 ug of purified vector were used for the cDNA from 4 ug of poly A+ RNA. The ligation reactions were composed as described for the adaptor addition above. The ligation reactions were transformed into MC1061/p3 cells made competent as described above. The transformation efficiency for supercoiled vector was 3-5x10⁸ colonies/ug.

Recovery and characterization of the CD28 clone.

[0140] Panning of the library was carried out as described herein above, using purified antibody 9.3 (DuPont) at a concentration of 1 ug/ml in the antibody cocktail. The methods used for COS cell transfection, radioimmunoprecipitation, RNA and DNA blot hybridization, and DNA sequencing were all as described herein above.

[0141] To isolate the CD27 cDNA, a large plasmic cDNA library was constructed in a high efficiency expression vector containing an SV40 origin of replication. A preferred version of the vector, containing an M13 origin, is shown in Figure 6. Three features of the vector make it particularly suitable for this use: (i) the eukaryotic transcription unit allows high level expression in COS cells of coding sequences placed under its control; (ii) The small size and particular arrange­ment of sequences in the plasmid permit high level replication in COS cells; and (iii) the presence of two identical BSTXI sites in inverted orientation and separated by a short replaceable fragment allows the use of an efficient oligonucleotide-based strategy to promote cDNA insertion in the vector.

[0142] The BstXI cleavage site, CCAN′₅NTGG, creates a four base 3′ extension which varies from site to site. A vector was created in which two identical sites were placed in inverted orientation with respect to each other, and separated by a short replaceable segment of DNA. Digestion with BstXI followed by removal of the replaceable segment yielded a vector molecule capable of ligating to fragments having the same ends as the replaceable segment, but not to itself. In parallel, cDNA synthetic oligonucleotides were employed that give the same termini as the replaceable segment. The cDNA then could not ligate to itself, but could ligate to the vector. In this way, both cDNA and vector were used as efficiently as possible.

[0143] Tailing with terminal transferase achieves the same end, but with less convenience and less overall efficiently. Moreover, homopolymer tracts located 5′ to cDNA inserts have been reported to inhibit expression in vitro and invivo (Yokota, T., et al., Nucl. Acids Res. 14:1511-1524 (1986); Riedel, H., EMBO J. 3:1477-1483 (1985)). Similar approaches based on the use of partially filled restriction sites to favor insertion of genomic DNAs (Zabarovsky, E.R., et al., Gene 42:119-123 (1986)) and cDNAs (Yang, Y., et al., Cell 47:3-10 (1986)) recently have been reported. These approaches give 2 or 3 base complementary termini, which usually ligate less efficiently than the 4 base extensions reported here.

[0144] Although the cloning scheme of the present invention does not result in a directional insertion of the cDNA, the ability to make large libraries easily, coupled with a powerful selection procedure, mades directional insertion unnecessary. The library construction efficiencies observed according to the present invention, between 0.5 and 2x10⁶ recombinants per ug of mRNA, with less than 1% background and an insert size greater than 1kb, compared favorably with those described for phage vectors lambda gt10 (7.5 x 10⁵d/ug of mRNA) and lambda gt11 (1.5 X 10⁶/ug of mRNA) (Huynh, T., et al.,, In: DNA Cloning Vol I, A Practical Approach, Glover, D.M. (ed.), IRL Press, Oxford (1985), pp. 49-78); but the resulting clones were more convenient to manipulate.

[0145] Surface antigen CDNAs can be isolated from these libraries using the antibody enrichment method of the present invention. In this method, the library is introduced into COS cells (for example, by spheroplast or protoplast fusion), where it replicates and expresses its inserts. The cells are harvested by detaching without trypsin, treated with monoclonal antibodies specific for the surface antigens desired, and distributed in dishes coated with affinity purified antibody to mouse immunoglobulins. Cells expressing surface antigen adhere, and the remaining cells can be washed away. From the adherent cells, a Hirt fraction is prepared (Hirt, B., J. Molec. Biol. 26:365-369 (1967)), and the resulting DNA transformed back into E. coli for further rounds of fusion and selection. Typically, after two rounds of selection with monoclonal antibodies recognizing different surface antigens, a single round of selection is performed with a single antibody, or pool of antibodies recognizing the same antigen.

Isolation of a CD28 cDNA.

[0146] The CD28 cDNA was isolated from a library of about 3 X 10⁵ recombinants prepared from cDNA from 0.8ug of poly A⁺ RNA using an earlier version of the protocol described in the Materials and Methods. The library was screened for CD28 (and other surface antigen) cDNA clones by the method outlined above. After the third transfection, COS cells were panned with the 9.3 antibody alone. A Hirt supernatant was prepared from the adherent cells and transformed into E. coli. Plasmid DNA was isolated form eight colonies and transfected individually into COS cell cultures. The presence of the DC28 antigen was detected in three of eight transfected cultures by indirect immunofluorescence. All three plasmid DNAs contained an insert of about 1.5 kb.

cDNA sequence analysis.

[0147] The CD28 cDNA encodes a long open reading frame of 220 residues having the typical features of an integral membrane protein (Figure 17). Removal of a predicted (von Heijne, Nucl. Acids Res. 14:4683-4690 (1986)) N-terminal signal sequence gives a mature protein of 202 residues comprising an extracellular domain with five potential N-linked glycosylation sites (Asn-X-Ser/Thr), a 27-amino acid hydrophobic membrane spanning domain, and a 41-amino acid cytoplasmic domain. Comparison of the amino acid sequence of CD28 with the National Biomedical Research Foundation database (Version 10.0) revealed substantial homology with mouse and rabbit immunoglobulin heavy-chain variable regions over a domain spanning almost the entire extracellular portion of CD28. Within this domain two cysteine residues in the homology blocks Leu-(Ser or Thr)-Cys and Tyr-(Tyr or Phe)-Cys are shared by CD28, CD4, CD8, immunoglobulin heavy- and light-chain variable sequences and related molecules with approximately the same spacing (Maddon et al., Annu. Rev. Biochem. 48:961-997 (1979)).

CD28 cDNA directs the production of a homodimer in transfected COS cells.

[0148] Immunoprecipitation of CD28 antigen from transfected COS cells was carried out using the monoclonal antibody 9.3 (Hansen, J.A., et al., Immunogenetics 10:247-260 (1980)). The material obtained from COS cells migrated with a molecular weight of 74kd under nonreducing conditions and 39 kd under reducing conditions, a pattern consistent with homodimer formation. Under the same conditions activated T cells give bands with molecular weights of 87 and 44kd, and HPB-ALL cells give bands of 92 and 50 kd, under nonreducing and reducing conditions respectively. The variation in molecular weight of the material obtained from different cell types arises as a result of differing glycosylation patterns characteristic of each type. Similar results were observed with other leukocyte surface antigens (Seed et al, Proc. Natl. Acad. Sci USA 87 (1987)). The nucleotide sequence of the CD28 cDNA predicts a mature protein with molecule weight of 23 kd, much smaller than observed in these experiments, and probably attributable to utilization of the 5 N-linked glycosylation sites predicted by the amino acid sequence.

RNA blot analysis.

[0149] Equal amounts of total RNA prepared from cell types expressing or lacking CS28 were subjected to RNA blot analysis as described hereinabove. Four bands with molecular weights of 3.7, 3.5, 1.5, and 1.3 kb were visible in lanes containing RNA thymocytes, T blasts, senescent T cells, and the T cell leukemia cell lines PEER and HPB-ALL. No bands were detected in lanes containing RNA prepared from the cell lines U937 (histiocytic leukemia), HuT-78 (Adult T cell leukemia), Jurkat (T cell leukemia), Namalwa (Burkitt lymphoma), MOLT4, and HSB-2, all of which do not express CD28. The 1.5 kb transcript presumably corresponds to the isolated cDNA, and the 3.7 and 3.5 kb species reflect incomplete splicing or alternative polyadenylation site utilization. The 1.3 kb transcript may terminate at an unconventional polyadenylation signal, since there is no obvious candidate in the sequence.

The CD28 gene is not rearranged.

[0150] DNA blot analysis (Seed et al., Proc. Natl. Acad. Sci USA 87 (1987)) of genomic DNA from placenta, peripheral blood lymphocytes, T cells, HeLa cells, or the tumor lines used in the RNA blot analysis above showed identical Dra 1 digest patterns indicating that rearrangement is not involved in the normal expression of the CD28 gene during development. Similarly, no gross genomic rearrangement underlies the failure of the examined T-cell tumor lines to express CD28 antigen. It may be inferred from the Dra 1 fragment pattern that the CD28 gene contains at least two introns.

EXAMPLE IV Isolation and Molecular Cloning of Two Human CD7 Antigen cDNAs

[0151] The CD7 cluster of antibodies (Palker, et al., Leukocyte Typing II, Springer-verlag, New York, 303-313 (1985)) recognized a 40 kd glycoprotein (gp40) on the surface of peripheral blood T cells and thymocytes. Early studies with anti-CD7 antibodies showed that CD7⁺ T cells enhance immunoglobulin (Ig) synthesis by B cells (Miroshima et al., J. Immunol. 129:1091-1098 1982)), suppress B cell Ig synthesis when stimulated with Concanavalin A (Haynes et al. Proc. Natl. Acad. Sci. U.S.A. 76:5928-5833 (1979)) and are the precursors of the cytotoxic T cells generated in mixed lymphocytic culture (Morishima et al., J. Immunol. 129:1091-1098 (1982)). Furthermore, CD7 has been found to be the most reliable marker for the identification of T cell acute lymphoblastic leukemia (Link et al., Blood 62:722-728 (1983)). As such, studies have been carried out, in which cytotoxins coupled to the anti-CD7 antibody 3A1 were used to purge bone marrow prior to reinfusion to avoid early relapse in autologous bone marrow transplants or as prophylaxis against graft vs. host disease in allogenic bone marrow transplants (Ramakrishnan et al., J. Immuol. 135:3616-3622 (1985)). Similarly, anti-CD7 antibodies also show promise as immunosuppressive agents in the treatment of allograft rejections (Raftery et al., Transpl. Proc., 17:2737-2739 (1985)) which is in accord with the recent observation that the anti-CD7 antibody 7G5 significantly inhibits the primary mixed lymphocyte reaction (Lazarovits et al. Leukocyte Typing III. Oxford Univ. Press, Oxford (1987)).

[0152] At present the physiological role of CD7 is not understood. It is known that anti-CD7 antibodies are not mitogenic, and do not block the T cells' response to PHA, or tetanus toxoid (Palker et al. Leukocyte Typing . Springer-verlag, New York, 303-313 (1985)). Some have noted that expression of CD7 in thymocytes occurs prior to the onset of T cell receptor beta-chain rearrangement (Pittaluga et al., Blood, 68:134-139 (1986)) and have pointed to a possible role for CD7 in this rearrangement and subsequent expression of the T cell receptor. It is clear that the cloning of the CD7 antigen would further efforts to understand its role in T cell physiology. Nucleotide sequencing and preliminary characterization of two cDNAs encoding the CD7 antigen was carried out according to the method of the present invention. Prompted by the recent suggestion that CD7 may be, or be part of, the T cell IgM receptor (Sandrin et al., Leukocyte Typing III. Oxford Univ. Press, Oxford (1987)), the ability of COS cells expressing CD7 to bind IgM or IgM immune complexes was evaluated. The results do not support the simple notion that CD7 itself is an IgM receptor.

Preparation of cDNA library and recovery and characterization of CD7 clones

[0153] Preparation of an HPB-ALL cDNA library in the expression vector piH3 was carried out as described herein. Panning of the library was carried out according to the method of the present invention, using purified anti-CD7 antibody Leu9 (Becton Dickinson) and antibody 7G5 as ascites fluid was diluted 1/1000. Methods for cell transfection, radioimmunoprecipitation, DNA and RNA blot hybridization and DNA sequencing were all as described herein.

IgM and IgG binding by COS cells transfected with CD7 and CDw32

[0154] Human Igm, IgG, and IgA antibodies, affinity purified FITC conjugated goat anti-human immunoglobulins antibodies (anti-Ig(G+M+A)), washed and preserved bovine red blood cells, and IgG and IgM fractions of rabbit anti-bovine red blood cell antibodies were purchased from Cooper Biomedical (Malverne, PA). COS cells were transfected by the DEAE Dextran method with cDNAs encoding the CD7, CDw32, and CD28 surface antigens. 48 hours after transfection the cells were washed with PBS/0.5% BSA and incubated with either human IgM, IgG or IgA antibodies at a concentration of 1ug/ml, at 4°C for 2 hours. Subsequently the cells were washed with PBX/0.5% BSA and incubated for 30 minutes at 4°C with FITC conjugated rabbit anti-human immunoglobulins. After washing the cells were examined with a fluorescence microscope. The experiments were also performed in the presence of 0.1% azide with the same results.

[0155] Bovine erythrocytes for rosette assays were prepared as described by Ercolani et al., J. Immunol. 127:2044-2051 (1981). Briefly, a 2% suspension of bovine erythrocytes was washed with PBS/0.5% BSA and treated with subagglutinating amounts of either IgG or the IgM fraction of rabbit anti-bovine erythrocyte antibodies at 4°C for 1 hour. Erythrocytes were then washed twice with PBS/0.5% BSA and adjusted to a 2% solution. 2 ml of antibody-coated erythrocytes were layered on 60mm dishes containing COS cells which has been transfected 48 hours earlier with either CD7, CD32 or CD28 by the DEAE Dextran method. The dishes were then centrifuged at 150 X g at 4°C for 15 minutes. After an additional 45 minute incubation at 4°C, the plates were gently washed 5 times with the 5 mls of PBS/0.5% BSA, and the COS cells were examined for rosette formation. These experiments were also performed in the presence of 0.1% sodium azide without alteration of the results.

Formation of T cell rosettes with antibody-coated erythrocytes

[0156] Peripheral blooc lymphocytes were obtained from heparinized blood by centrifugation at 4°C over a Ficol-Hypaque gradient at 400 x g for 30 minutes. Leukocytes at the interface were washed two times with PBS. The leukocytes were adjusted to 10Y7 cells/ml in IMDM/10% Fetal Bovine Serum (FBS) and incubated in tissue culture dishes at 37°C for 30 minutes. Nonadherent cells were transferred to new dishes, and PHA was added to stimulate proliferation of T lymphocytes. On the next day the cells were washed with PBS and placed in fresh IMDM/10%FBS.

[0157] Rosette assays were performed three days later. Cells were washed with PB/0.5% BSA, and a 10 ul suspension of 2% Ig-coated erythrocytes prepared as described above was added to 10 ul of PBS/0.5% BSA containing 5 X 10⁶ cells/ml. The mixtures were placed in Falcon round bottom 96 well plates and centrifuged at 150 X g for 15 min at 4°C. After an additional incubation of 45 min at 4°C pellets were resuspended with 10 ul of PBS/0.5% BSA, and the rosettes scored by phase contrast microscopy. The experiments were carried out in both the presence and absence of 0.1% sodium azide with no detectable difference.

Isolation of cDNAs encoding the human CD7 antigen.

[0158] To isolate CD7 cDNAs, a large plasmid library was constructed in the expression vector H3M as describe hereinabove. The library was introduced into COS cells by spheroplast fusion, and allowed to replicate and express its inserts. The COS cells were harvested by detaching without trypsin 48 to 72 hours after transfection, treated with monoclonal antibodies specific for surface antigens believed to be encoded in the library, and distributed in dishes coated with affinity purified anti-mouse antibody as described herein. Under these conditions, cells expressing surface antigen adhere and the remaining cells can be washed away.

[0159] A Hirt (Hirt, J. Mol. Biol., 26:365-369 (1967)) fraction was prepared from adherent cells, and the resulting DNA transformed back into E. coli for further rounds of fusion and selection. In the third round of selection the detached cells were treated with a mixture of monoclonal antibodies specific for CD7 (765 and Leu9), and a Hirt supernatant was again generated and transformed into E. coli. After transformation of the DNA into E. coli 8 colonies were picked, and the plasmid DNA prepared from them by an alkaline miniprep procedure (Maniatis, et al., Molelcular Cloning: A Laboratory Manual, Cold Spring Harbor Press, Cold Spring Harbor, New York (1982)). DNA was prepared from 8 resulting colonies and transfected into COS cells. After 3 days, surface expression of the CD7 antigen was detected by indirect immunofluorescence in 7 of 8 transfected dishes. Restriction enzyme digest of the corresponding plasmid DNAs revealed two species. One contained a 1.2 kb insert, and the other a 1.3 kb insert.

CD7 cDNA sequence analysis

[0160] Both isolates were sequenced by the dideoxynucleotide method. The 1.2 kb cDNA encodes a long open reading frame of 240 residues having the typical features of an integral membrane protein. The initial assignment of the signal sequence cleavage site by the method of von Heijne (Nucl. Acids Res. 14:4683-4690 (1986)) was at the 18th residue. It later was determined, however, that the homology with immunoglobulin variable regions would better predict the mature terminus at residue 26; this assignment would also correlate well with the position of the intron as discussed below and as shown in Figure 8. Removal of the predicted N-terminal signal sequence gives a mature protein of 215 residues with a predicted molecular mass of 23kd. In the extracellular domain are two N-linked glycosylation sites (Asn-X-Ser Thr), in agreement with the results of Sutherland et al. (J. Immunol. 133:327-333 (1984)), who also showed the presence of O-linked glycans and covalently associated palmitic acid on the mature protein. In the 27 amino acid hydrophobic membrane spanning domain is a single cysteine residue which may be the site of fatty acylation (Rose et al., Proc. Natl. Acad. Sci. USA 81:2050-2054 (1984); Kaufman et al., J. Biol. Chem. 259:7230-7238 (1984)). The length of the cytoplasmic domain, 39 residues,is in good agreement with the 30-40 amino acids predicted by protease digestion of the CD7 precursor in rough microsomal membrane fractions (Sutherland et al., J. Immunol. 133:327-333 (1984)).

[0161] Sequence analysis of the 1.7 kb clone (Figure 8) revealed the presence of an intron located 121 bp from the 5′ end. The 411 bp intron contains stop codons in all three reading frames and is located just downstream of the secretory signal sequence, as is frequently observed for secreted or surface proteins. Both the 5′ and 3′ ends of the intron conform to the splice donor/acceptor consensus AAG GTRAGA/.../Y₆₋₁₁NYAG A (Mount, Nucl. Acids Res. 10:459-472 (1982)). Because both the 1.2 and 1.7 kb clones express CD7 antigen equally well in COS cells, the intron must be excised in COS cells fairly efficiently.

[0162] Comparison of the amino acid sequence with the National Biomedical Research Foundation database revealed substantial homology with human and mouse immunoglobulin kappa chain and T-cell receptor gamma chain variable regions over almost the entire extracellular portion of the molecule. Two cysteine residues shared in approximately equal spacing by all three structures fall in the conserved sequences Ile-Thr-Cys and Tyr-X-Cys. In kappa chain variable regions these cysteins form a disulfide bridge. The presence of at least one intrastrand disulfide bond in the CD7 structure has previously been proposed by Sutherland et al. (J. Immunol. 133:327-333 (1984)), who noted that immunoprecipitation of CD7 gave rise to a band with an apparent molecular mass of 40 kd under reducing conditions and 38 kd under nonreducing conditions.

[0163] Based on the homology with immunoglobulin V-regions, it is predicted that CD7 contains a disulfide bond linking Cys 23 and Cys 89. A second disulfide bond, linking Cys 10 and Cys 117, has been proposed, based on the structural similarity between CD7 and Thy-1. The extracellular domains of both Thy-1 and CD7 have 4 cysteine residues, in roughly homologous positions. The 4 cysteine residues, in roughly homologous positions. The 4 cysteine residues of Thy-1 are joined in two internal disulfide bridges between Cys 9-111 and Cys 19-85 (Williams et al., Science 216:696-703 (1982)). In Thy-1, Cys 111 forms an amide bond with the ethanolamine moiety of a substituted phosphatidylinositol, and is thus the last residue of the mature moleculs (Tse et al., Science 230:1003-1008 (985)). In CD7, Cys 117 is followed by four repeats of a sequence whose consensus is Xaa-Pro-Pro-Xaa-Ala-Ser-Ala-Leu-Pro, and which, it is proposed, plays the role of a stalk projecting the V-like domain away from the surface of the cell.

[0164] In addition to the homologies shown in Figure 20 and mentioned above, the extracellular domain of CD7 has significant homology with both chains of the rat CD8 heterodimer (Johnson et al., Nature 323:74-76 (1986)), and the myelin P₀ protein (Lemke et al., Cell 40:501-508 (1985)).

CD7 directs the production of a 40 kd protein in transfected COS cells

[0165] Immunoprecipitation of CD7 antigen from transfected COS cells was carried out as described herein using monoclonal antibody 7G5 (Lazarovits et al., Leukocyte Typing III, Oxford Univ. Press, publisher, Oxford, England (1987). The material obtained from COS cells migrated with as a broad band with molecular weight of 40 kd under reducing conditions. Under the same conditions HPB-All cells (the cDNA donor line) and activated T cells gave bands with molecular widths of 41 and 39 kd respectively. In both the COS cell and HPB-ALL lane a faint band with molecular weight of 30 kd was also observed, possibly corresponding to a partially glycosylated precursor (Sutherland, D.R., et al., J. Immunol. 133:327-333 (1984)).

RNA blot analysis

[0166] Equal amounts of total RNA prepared from cell types expressing or lacking CD7 were subjected to Northern blot analysis as described herein. A single 1.3 kb species was visible in lanes containing RNA from thymocytes, activated T cells, resting T cells, and the T cell leukemia lines HuT-78, HPB-ALL, Jurkat J3R7, HSB-2 and PEER. With the exception of the PEER cell line, none of the T cell tumors showed significant overexpression of CD7 transcripts. CD7 RNA was detected in all of the thymus-derived cells, but not in RNA from U937 (histiocytic leukemia) and Namalwa (Burkitt Lymphoma) cells. No band corresponding to the 1.7 kb cDNA could be detected, suggesting that this species is artificially enriched during the cloning or library amplification process.

[0167] Enrichment during amplification seems unlikely because the 12 kb cDNA clone propagates as well in E. coli as the 1.7 kb clone. However, immediately upstream and downstream from the site of insertion of the intron are sequences that could form an interrupted stem and loop structure. Eight of the 10 basepairs of the potential stem at GC pairs, perhaps giving the structure sufficient stability to interfere with elongation of the cDNA first strand. The presence of the intron greatly separates the two halves of the stem, potentially eliminating the structure via unfavorable loop entropy and allowing efficient first strand synthesis.

The CD7 gene is not rearranged

[0168] Southern blot analysis of genomic DNA from placenta, peripheral blood lymphocytes, T cells, HeLa cells, or the tumor lines used in the RNA blot analysis above showed identical Dra 1 digest patterns. Thus, the CD7 gene is not grossly altered during development, and the high level of expression in the PEER cell line is not the consequence of a substantial genomic rearrangement.

COS cells expressing CD7 do not bind IgM

[0169] Human peripheral blooc T lymphocytes express receptors for IgM antibodies (FcRu: Moretta et al., Eur. J. Immunol. 5:565-569 (1975); McConnell et al., Immunol. 30:835-837 (1976)). Recently it has been reported that CD7 might play a role in IgM binding by T cells (Sandrin et al., Leukocyte Typing III, Oxford Univ. Press, publisher, Oxford, England (1987)). L cells, normally CD7⁻ and FcRu⁻, become CD7⁺ and FcRu⁺ when transfected with a 16 kb genomic fragment encoding the CD7 antigen (Sandrin et al., Leukocyte Typing III, Oxford, Univ. Press, publisher, Oxford, England (1987)). Furthermore, IgM binding to CD7-positive cells can be blocked by the anti-CD7 monoclonal antibody Huly-m2 (Thurlow et al, Transplantation 38:143-147 (1984)), and IgM columns bind a 37 kd protein from radiolabeled lysates of peripheral blood T lymphocytes (Sandrin et al., Leukocyte Typing III, Oxford Univ. Press, publisher, Oxford, England (1987)).

[0170] Accordingly, COS cells expressing CD7 were tested for their ability to bind IgM. IgM receptor activity was assayed either by direct binding (Hardin et al., Proc. Natl. Acad. Sci. USA 76:912-914 (1979)) or by a rosette assay with ox erythrocytes coated with an IgM fraction of rabbit anti-bovine red cell serum as described by Ercolani et al. J. Immunol. 127:2044-2051 (1981)). Cells expressing CD7 neither bound human IgM nor formed rosettes with Igm-coated erythrocytes. Under the same conditions, COS cells transfected with a cDNA encoding the human IgG receptor CDw32 bound IgG directly and formed rosettes with IgG-coated erythrocytes. Erythrocytes coated with IgM or IgG antibodies also adhered to a fraction of peripheral blood lymphocytes as reported (Moretta et al., Eur. J. Immunol. 5:565-569 (1975)).

[0171] These results do not support the notion that the CD7 antigen is by itself an IgM receptor, although they do not exclude the possibility that COS cells suppress IgM binding activity in some manner, or that CD7 is part of, or modified to become, an IgM receptor. That CD7 is not by itself an IgM receptor is supported by the observation that a number of CD7⁺ T cell lines are FcRu-(Sandrin et al., Leukocyte Typing III, Oxford, Univ. Press, publisher, Oxford, England (1987)).

EXAMPLE V Isolation and Molecular Cloning of the Human CDw32 Antigen

[0172] A cDNA encoding the human CDw32 antigen, a human receptor for immunoglobulin G constant domains (Fc receptor), was isolated by the method of the present invention, by virtue of its affinity for its ligand, IgG. The sequence of the isolated clone is most closely related to the murine beta 2 Fc receptor, but has diverged completely in the portion encoding the cytoplasmic domain. The receptor expressed in COS cells shows a preference for IgG₁ among IgG subtypes, and no affinity for IgM, IgA or IgE.

[0173] To isolate the Fc receptor clone, cDNA libraries were prepared from tumor cell lines or from a human tumor and transfected into COS cells. After 48 hours, the cells were treated with mouse or human IgG antibodies, and allowed to settle on dishes coated with affinity-purified sheep anti-mouse IgG or goat anti-human IgG antibodies. After lysis, DNA recovery, and transformation in E. coli, the cycle was repeated for two more rounds. Although no positive clones were isolated from the tumor line libraries, a cDNA clone encoding an Fc receptor was isolated from a library prepared from a human adrenal tumor. It has been discovered that many tumors are heavily infiltrated by macrophages and lymphocytes. Thus, tumor RNA may be a productive source in general for transcripts of human macrophages.

[0174] By indirect immunofluorescence assay, the human receptor expressed on COS cells bound all mouse and human IgGs with relatively low affinity -10⁻⁷M), and a clear discrimination was noted among human antibodies for IgG₁. Human IgM, IgA₁, and IgA₂, and IgE did not bind, nor did murine IgM or IgA. As expected, human Fc, but not Fab fragments, bound to the transfected cells. Among monoclonal antibodies donated to the Third International Workshop on Leukocyte Differentiation Antigens, three gave strong positive immunofluorescence: two (out of two) recognizing the Fc Receptor CDw32 determinant, and one (out of four) recognizing the CD23 (B cell IgE Fc receptor) determinant. Monoclonals recognizing the T cell/Macrophage Fc receptor antigen CD16 gave only weak immunofluorescence comparable to that shown by control ascites.

[0175] Radioimmunoprecipitation of transfected COS cells with CDw32 antibodies showed the presence of a single 40kd species, comparable in size to the antigen recognized on the surface of the myeloid CDw32⁺ line HL-60, and to the less abundant antigen present on the histiocytic leukemia line U937. This result reinforces the notion that the isolated receptor is CDw32, as the CD16 receptor is reported to be substantially larger (60-70kd).

[0176] The nucleotide sequence of the isolated receptor (Figure 9) is highly homologous to that of members of the recently isolated murine receptor family, and most closely related to the murine beta₂ receptor by nucleic acid homology. Surprisingly, the murine beta₂ receptor is found on T and B lymphocytes and macrophages, while the alpha receptor is restricted to macrophages; in the human system, CDw32 (shown here to be beta₂-like) is restricted to macrophages while another Fc receptor (CD16) is found on lymphocytes and macrophages. The human sequence appears to have diverged from the mouse sequence by insertion of approximately 1kb of DNA a few bases 3′ to the junction between the transmembrane and cytoplasmic domains. The junctions of the insertion site do not show obvious relationships to splice donor and acceptor sequences. Comparison of the human and murine peptide sequences showed that the peptide sequence diverges at the end of the transmembrane domain, before the nucleotide sequence diverges, suggesting the existence of a selective pressure favoring the creation of a different cytoplasmic domain.

[0177] RNA blot analysis showed that myeloid but not lymphocytic cell lines expressed RNA homologous to the CDw32 probe. DNA blot analysis showed multiple bands consistent with the existence of a small multigene family.

EXAMPLE VI Isolation and Molecular Cloning of Two cDNA Clones Encoding the B Lymphocyte-­specific CD20 (B1, Bp35) Antigen

[0178] Recent studies suggest that the pan B cell antigen CD20 (B1, Bp35) plays an important role in B cell activation. Monoclonal antibodies (mAb) to CD20 induce different cellular responses depending on the antibody used and the stage of differentiation or activation of the target B cells. The monoclonal antibody IF5 activates resting B cells by initiating the transition from the G₀ to the G₁ phase of the cell cycle, and induces dense tonsillar B cells to proliferate (Clark et al., Proc. Natl. Acad. Sci USA 82:1766 (1985); Clark and Shu, J. Immunol. 138:720 (1987)). However, 1F5 does not induce an increase in cytoplasmic free calcium and does not induce circulating B cells to proliferate (Rabinovitch et al., In: Leukocyte Typing III (McMichael, Ed.), p. 435, Oxford University Press (1987))/ Other anti-Cd20 mAbs, such as B1, have been shown to block B cell activation (Tedder et al., J. Immunol. 135:973 (1985)) and both 1F5 and B1 can inhibit B cell differentiation (Golay et al., J. Immunol. 135:3795 (1985)). Recently it has been suggested that phosphorylation and internalization of CD20 may be necessary steps for B cell entry into the G₁ phase of the cell cycle (Valentine et al., In: Leukocyte Typing III (McMichael, Ed.), p. 440, Oxford University Press (1987)). In the present example, two CD20 cDNA clones were isolated and expressed using the methods of the present invention.

Preparation of cDNA Library and Recovery of cDNA Clones by Panning

[0179] Poly (A)⁺ RNA was prepared from the human Burkitt cell line Daudi by oligo (dT) cellulose chromatography of total RNA isolated by procedures described herein. cDNA preparation and expression library construction were carried out as described.

[0180] Anti CD20 mAbs 1F5, 2H7, B1, L27, G28-2, 93-1B3, B-C1, and NU-B2 were obtained from the International Leukocyte Typing Workshop (Valentine et al., In: Leukocyte Typing III (McMichael, Ed.), p. 440, Oxford University Press (1987)). Purified mAbs were used at a concentration of 1ug/ml and ascites were used at a dilution of 1:1000. Panning was done according to the present method. In the first round of screening, eight 10cm dishes of 50% confluent COS cells were transfected by the DEAE-Dextran method. Subsequent screening cycles were performed by spheroplast fusion.

Immunoprecipitation, Sequencing, RNA and DNA Blot Hybridization

[0181] B cell lines CESS and Daudi were metabolically labeled with ³⁵S-methionine and ³⁵S-cysteine for 6h at 37°C. COS cells transfected by the DEAE-Dextran method were similarly labeled 36 hours post-transfection. The labeled cells were incubated with the B1 mAb (Coulter) at 4°C for 1 h, washed in PBS, and lysed with 0.5% NP-40, 0.1% SDS, 0.5% deoxycholate and 1mM PMSF in PBS. After centrifuging (13000xg, 5 min.), the lysate was incubated with fixed S. aureus cells (Calbiochem) for 1 hr at 4°C. The S.aureus cells were pelleted, washed 5 times with 1% NP-40/PBS, eluted and electrophoresed through 12.5% polyacrylamide gels.

[0182] DNA and RNA blot analysis and hybridization probe preparation were carried out as described. Sequencing was done by the method of Sanger et al. Proc. Nat. Acad. Sci. USA 74:5463 (1977). The nucleotide sequence of the CD20.4 cDNA is represented in Figure 10.

[0183] Two cDNA clones, bearing inserts of 1.5 (CD20.4) and 1.0 kb (CD20.6), were isolated from a Daudi cell DNA library by panning with a panel of mAbs against CD20. COS cells transfected with either clone reacted with all members of the panel of antibodies. Immunoprecipitation of the cDNA-encloded protein from transfected COS cells showed two bands of 32 and 30kd reminiscent of the 37 and 35kd bands observed in different B cell subsets and lines (Valentine et al., "Structure and Function of the B Cell Specific 35-37kDa CD20 Protein," In: Leukocyte Typing III, A. McMichael et al., Oxford University Press, p. 440 (1987)). It has been the experience of the present inventors that the molecular masses of surface antigens expressed in COS cells are consistently smaller than those of their native counterparts. This may be due to differences in glycosylation.

[0184] Both cDNA clones have the same coding sequence, and differ only in the 3′ untranslated region. The insert in clone CD20.6 has a short polyA tail and lacks a consensus polyadenylation signal, while the insert in CD20.4 lacks a polyA tail and extends 431 bp beyond the 3′ terminus in CD20.6 (Fig. 10A).

[0185] RNA blot analysis showed that three transcripts of 3.8, 3.0 and 1.5 kb were present in B cells but absent from other cell types, in agreement with the known pattern of antibody reactivity (Clark et al., Proc. Natl. Acad. Sci. USA 82:1766 (1985); Clark et al, J. Immunol. 138:720 (1987); Tedder et al., J. Immunol. 135:973 (1985); Golay et al., J. Immunol. 135:3795 (1985)). It appears likely that the Cd20.6 clone is derived from the 1.5 kb transcript or possibly from an even shorter, undetectable species. Because the CD20.4 clone lacks a poly(A)⁺ tail, its source cannot be inferred at present.

[0186] DNA blot analysis showed that the CD20 genomic sequences are not rearranged during development and are not amplified in the cell lines examined. A restriction fragment length polymorphism was observed in a DNA sample obtained from placenta.

[0187] The amino acid sequence predicted by the cDNA contains 297 residues and has a molecular mass of 33,097 daltons. The sequence contains three major hydrophobic stretches involving residues 51-103, 117-141 and 183-203 (Fig.10). Two other notable characteristics are the absence of an amino-terminal signal peptide and the presence of a highly charged carboxy-terminal domain. A polyclonal anti-CD20 antibody that recognized the last 18 residues of the carboxy-terminus reacts with lysates of cells expressing CD20 but not with intact cells, suggesting that the CD20 carboxy terminus is located within the cytoplasm. Since there is no amino-terminal signal peptide, it is likely that the amino-terminus is also intracellular, and that the first hydrophobic region acts as an internal membrane insertion signal (Zerial et al., EMBO J. 5:1543 (1986)). The first hydrophobic region is composed of 53 residues and is therefore long enough to span the membrane twice if organized as an alpha helix. Because there are two remaining hydrophobic regions, the intracellular side. Alternatively, the carboxy-terminus requires that the first hydrophobic domain exit the membrane on the intracellular side. Alternatively, the carboxy-terminal antibody may only recognize epitopes exposed by detergent treatment allowing the carboxy-terminus to be extracellular and forcing the first hydrophobic domain to exit the membrane on the extracellular side. The sequence contains 2 potential N-glycosylation sites (Asn-Xaa-Ser/Thr, where Xaa cannot be Pro (Bause, Biochem. J. 209:331 (1983)) at positions 9 and 293, but neither of these is expected to be used if located in intracellular domains of the molecule. The difference in molecular mass between CD20 expressed on COS cells and on B cells is therefore presumably due to 0-linked glycosylation, although other forms of post-translational modification are not excluded. If the carboxy-terminus is intracellular, the only extracellular domain would lie between residues 142 and 182. This region is rich in serine and threonine residues which might support 0-glycosylation.

[0188] The observation of two protein species in COS cells cannot be explained by alternate splice formation because the cDNA sequence does not contain any promising splice donor or acceptor sequences (Shapiro et al., Nucl. Acids Res. 15:7155 (1987)). A difference in glycosylation or alternate translational initiation site selection may account for the two species observed. Initiation at either the first or the second ATG gives protein molecular masses of 33.1 and 30.8 kd respectively, in good agreement with the sizes observed in COS cells. Neither ATG is embedded in the consensus sequence proposed by Kozak (Nucl. Acids Res. 12:857 (1984)). Use of alternate initiation sites has been reported for several proteins (Kozak, Nucl. Acids Res. 12:857 (1984)).

[0189] Comparison of the peptide sequence with the sequences in the National Biomedical Research Foundation database showed no significant homology by the FASTP rapid sequence alignment algorithm. Because the bulk of the protein appears to be confined to the interior of the membrane and the cell, it seems plausible that it may play a role in transducing signals from other transmembrane proteins to the cell interior. Consistent with this role is the relatively hydrophilic nature of the hydrophobic regions which might allow hydrogen bond interactions with the transmembrane portions of other proteins.

EXAMPLE VII Isolation and Molecular Cloning of ICAM, An Adhesion Ligand of LFA-1

[0190] Antigen-specific cell contacts in the immune system are strengthened by antigen-non-specific interactions mediated in part by lymphocyte function associated or LFA antigens (Springer, T.A., et al., Annu. Rev. Immunol. 5:223-252 (1987); Anderson, D.C., et al., Annu. Rev. Medicine 5:175-194 (1987)). The LFA-1 antigen, a major receptor of T cells, B cells and granulocytes (Rothlein, R., et al., Exp. Med. 163:1132-1149 (1987)), is involved in cytolytic conjugate formation, antibody-dependent killing by NK cells and granulocytes, and helper T cell interactions. LFA-1 has been placed in the integrin family of cell surface receptors by virtue of the high sequence similarity between the LFA-1 and integrin beta chains (Kishimoto, T.K., et al., Cell 48:681-690 (1987); Hynes, R.O. Cell 48:549-554 (1987)). The adhesion ligands of the integrin family are glycoproteins bearing the Arg-Gly-Asp (RGD) sequence motif, e.g., fibronectin, fibrinogen, vitronectin and von Willebrand factor (Ruoslahti, E., et al., Cell 44:517-518 (1987)).

[0191] In this example, the Intercellular Adhesion Molecule-1 (ICAM-1), a ligand for LFA-1 (Rothlein, R., et al., J. Immunol. 137:1270-1275 (1986); Dustin, M.L., et al. J. Immunol. 137:245-254 (1986)), was cloned according to the methods of the present invention. ICAM contains no RGD motifs, and instead is homologous to the neural cell adhesion molecule NCAM (Cunningham, B.A., et al. Science 236:799-806 (1987); Barthels, D., et al., EMBO J. 6:907-914 (1987)). COS cells transfected with the ICAM cDNA clone bind myeloid cells by a specific interaction which can be blocked by monoclonal antibodies directed against either LFA-1 or ICAM-1.

[0192] A cDNA library was constructed using RNA prepared from HL60 cells induced with phorbol myristyl acetate (PMA). The library was transfected into COS cells and cells expressing surface antigens were recovered according to the methods of the present invention by panning with the anti-ICAM monoclonal antibodies (mAbs) 8F5 and 84H10 (McMichael, A.J., et al., eds., Leukocyte Typing III. White Cell Differentiation Antigens, Oxford University Press (1987)). Episomal DNA was recovered from the panned cells and the expression-panning cycle repeated a further 2 times to obtain a cDNA clone designated pICAM-1.

[0193] COS cells transfected with pICAM-1 gave positive surface immunofluorescence reactions with three anti-ICAM-1 antibodies: 8F5; 84H10; and RR-1. Immunoprecipitation of pICAM-1-transfected COS cells with the mAb 84H10 gave a band of molecular mass 100kD. 30). A slightly larger protein of 110 kD was precipitated from HL60 cells induced for 48 hours with either phorbol myristyl acetate (PMA), gamma-interferon (gammaIFN), tumor necrosis factor (TNF), or interleukin-1 beta (IL-1 beta), but was absent from uninduced cells. The smaller molecular mass of ICAM-1 expressed in COS cells is consistent with the lower molecular masses observed for other surface antigens expressed in COS cells.

[0194] RNA blot analysis showed 2 species of 3.2 kb and 1.9 kb present in HL60 cells stimulated with either PMA, gamma IFN, TNF or IL-1 gamma, but absent in uninduced cells. Thus, the expression of ICAM-1 is regulated by a number of cytokines, apparently at the level of transcription. Similar species were present in B cells (JY and Raji), T Cells (Peer and T blasts) and Lymphokine Activated Killer Cells ′LAK). The structure of these ICAM-1 transcripts and their relationship to the pICAM-1 cDNA remains to be established. Blot hybridization of genomic DNA from placenta revealed a pattern consistent with a single copy gene.

[0195] To investigate whether pICAM-1 encodes a functional cell adhesion molecule, COS cells expressing ICAM-1 were tested for their ability to bind HL60 cells. After 30 minutes at 37°C in the presence of Mg²⁺, HL60 cells strongly adhered to the ICAM-expressing COS cells, but not to mock transfected cells. The specificity of this adhesion was demonstrated by preincubating the ICAM-1 expressing COS cells with mAb 84H10. All HL60 binding was abolished under these conditions. An isotype matched monoclonal antibody, W6/32, which recognizes a monomorphic HLA-ABC related determinant of approximately equal abundance to ICAM-1 on transfected COS cells, had no effect on the adhesion. Similarly, preincubation of the HL60 cells with either 84H10 or W6/32 did not inhibit binding.

[0196] To determine if LFA-1 was acting as the receptor for ICAM-1 in this system, HL60 cells were pretreated with antibodies against the beta chain of LFA-1 (CD18 (McMichael, A.J., et al., eds., Leukocyte Typing III. White Cell Differentiation Antigens, Oxford University Press (1987))) and then subjected to the binding assay. All adhesion to ICAM-expressing COS cells was blocked. Pretreatment of COS cells with the CD18 antibodies had no effect on the adhesion. This provides direct evidence that ICAM-1 is indeed acting as an adhesion ligand for LFA-1.

[0197] The sequence of the pICAM-1 cDNA insert consists of 1846 nucleotides (Fig. 11). The predicted peptide sequence of 532 residues has the typical features of a transmembrane protein including a putative signal sequence, which may be cleaved between glycine-25 and asparagine-26 (von Heijne, G., Nucl. Acids Res. 14:4683-4690 (1986)), and a single 25 residue membrane-spanning domain terminating in a short, highly charged cytoplasmic domain. The extracellular domain contains seven potential N-linked glycosylation sites which could adequately explain the difference in size between the deglycosylated precursor (55 kD) and the final product (90-115 kD) (Dustin, M.L., et al., J. Immunol. 137:245-254 (1986)). Differential use of these putative glycosylation sites could also explain the heterogeneous molecular mass of ICAm-1 observed in different cell types (Dustin, M.L., et al., J. Immunol. 137:245-254 (1986)).

[0198] LFA-1 is a member of the integrin family of cell surface receptors (Kishimoto, T.K., et al., Cell 48:681-690 (1987); Hynes, R.O., Cell 48:549-554 (1987)). The tripeptide motif Arg-Gly-Asp (RGD) is a common feature of the ligands for this family, e.g., fibronectin, fibrinogen, vitronectin and von Willebrand factor, and is crucial for ligand-receptor interaction (Ruoslahti, E, et al., Cell 44:517-518 (1987)). However, ICAM-1 contains no RGD motifs, bearing instead a single RGE sequence at position 152. A search of the National Biomedical Research Foundation (Dayhoff, M.O., et al., Methods Enzymol. 91:524-545 (1983)) (NBRF) database revealed no significant similarities to other proteins. However, a comparison to a laboratory database containing recently published surface proteins did reveal a surprising and significant similarity between ICAM-1 and the neural cell adhesion molecule NCAM-1 (Cunningham, B.A., et al., Scienc 236:799-806 (1987); Barthels, C., et al., EMBO J. 6:907-914 (1987)). The optical alignment score obtained using the NBRF ALIGN program is 8 standard deviations above the mean score obtained from 500 random permutations of the sequences. The probability of the spontaneous occurence of an equal or higher score is approximately 10⁻⁹.

[0199] Using a database of known immunoglobulin related sequences, it has been shown that ICAM-1 may be divided into five Ig domains (28-112, 115-206, 217-310, 312-391, and 399-477) each of which shows significant similarity with other members of the Ig superfamily (Williams, A.F., Immunol. Today 8:298-3-3 (1987)). For example, domain I is similar to CD3 whilst domains IV and V are similar to domains of myelin associated glycoprotein (Arguint, M., et al., Proc. Natl. Acad. Sci. USA 84:600-604 (1987)) and carcinoembryonic antigen (Beauchemin, N., et al., Mol. Cell. Biol. 7:3221-3230 (1987)). All five Ig domains of NCAM align with the Ig segments in ICAM, and the principal contribution to the similarity comes from domains II and III of ICAM. Finally, the T cell-specific adhesion molecule CD2 shows roughly the same similarity to NCAM as does ICAM, but ICAM and CD2 are only weakly related. Thus, some precursor of NCAM is ancestral to both ICAM and CD2.

[0200] The availability of a functional ICAM-1 cDNA will allow a better assessment of the role of ICAM-1/LFA-1 mediated adhesion in antigen-specific leukocyte function, including T-cell mediated killing, T-helper responses and antibody-dependent cell mediated killing.

EXAMPLE VIII Isolation and Molecular Cloning of the Human CD19, CD20, CD22, CDw32a, CDw32b, and CD40 Antigens

[0201] The rapid immunoselection cloning method of the present invention was applied to isolate and clone the CD19, CD20, CD22, CDw32a, CDw32b, and CD40 antigens. The nucleotide sequence of CD19 is shown in Figure 12. The nucleotide sequence of CD20 is shown in Figure 13. The nucleotide sequence of CD22 is shown in Figure 14. The nucleotide sequence of CDw32a is shown in Figure 15. The nucleotide sequence of CDw32b is shown in Figure 16. The nucleotide sequence of CD40 is shown in Figure 17.

Equivalents

[0202] Those skilled in the art will recognize or be able to ascertain, using no more than routine experimentation, many equivalents to the specific embodiments of the invention described herein. Such equivalents are intended to be encompassed within the scope of this invention.

Claims

1. A method for cloning a cDNA encoding a cell surface antigen, comprising

(a) preparing a cDNA library;

(b) introducing said cDNA library into mammalian tissue culture cells;

(c) culturing said cells under conditions allowing expression of said cell surface antigen by said cells;

(d) exposing said cells to a first antibody directed against said cell surface antigen selected from the group consisting of CD1a, CD1b, CD1c, CD6, CD13, CD14, CD16, CD19, CD20, CD22, CD26, CD27, CD31, CDw32a, CDw32b, CD33, CD34, CD36, CD37, CD38, CD39, CD40, CD43, FcRIa, FcRIb, TLiSa, and Lev-8;

(e) thereafter exposing said cells to a substrate coated with a second antibody directed against said first antibody, thereby causing said cells to adhere to said substrate; and

(f) separating adherent from non-adherent cells.

2. The method of claim 1 wherein said cDNA library is prepared from RNA extracted from a human cell line, preferably HPB-ALL or JY, or from RNA extracted from RNA extracted from human tumor cells, preferably human adrenal tumor cells.

3. A cloned cDNA segment encoding a cell surface antigen selected from the group consisting of CD1a, CD1b, CD1c, CD6, CD13,CD14, CD16, CD19, CD20, CD22, CD26, CD27, CD31, CDw32a, CDw32b, CD33, CD34, CD36, CD37, CD38, CD39, CD40, CD43, FcRIa, FcRIb, TLiSa, and Lev-8, or a functional derivative thereof.

4. A substantially pure protein selected fromt the group consisting of CD1a, CD1b, CD1c, CD6, CD13, CD14, CD16, CD19, CD20, CD22, CD26, CD27, CD31, CDw32a, CDw32b, CD33, CD34, CD36, CD37, CD38, CD39, CD40, CD43, FcRIa, FcRIb, TLiSa, and Lev-8, or a functional derivative thereof.

5. Substantially pure CDw32a, CDw32b, CD20, CD19, CD22 or CD40 antigen, or a functional derivative of these antigens.

6. Substantially pure cDNA comprising the nucleotide sequence shown in Figure 15, Figure 17, Figure 13, Figure 12, Figure 14, or in Figure 17, or a functional derivative of these cDNAs.

7. Substantially pure protein having the amino acid sequence shown in Figure 10, or a functional derivative thereof.

8. A pharmaceutical composition comprising substantially pure cell surface antigen, especially CD1a, CD1b, CD1c, CD6, CD13, CD14, CD16, CD19, CD20, CD22, CD26, CD27, CD31, CDw32a, CDw32b, CD33, CD34, CD36, CD37, CD38, CD39, CD40, CD43, FcRIa, FcRIb, TLiSa, or Lev-8, or a functional derivative of said cell surface antigen; said cell surface antigen present in soluble form or in the form of an acceptable salt, preferably additionally comprising an acceptable carrier and/or a therapeutic agent.

9. An immunotherapeutic agent, comprising substantially pure cell surface protein, an acceptable carrier and a therapeutic agent.

10. A kit useful for immunodiagnostic assay, comprising substantially pure cell surface antigen in soluble form and a containing means.

Drawing